The Carbon Robotics LaserWeeder smart implement can autonomously remove unwanted plants from the crop field. | Credit: Carbon Robotics

Carbon Robotics closed $30 million in Series C financing from new lead investor Sozo Ventures along with existing investors Anthos Capital, Fuse Venture Capital, Ignition Partners, Liquid2 and Voyager Capital. The funding will be used to expand sales regions in North America, optimize and scale manufacturing, develop new software and hardware products, and launch into international markets. This latest round of financing brings Carbon Robotics’ overall funding to $67 million.

“Carbon Robotics’ elegant use of AI, computer vision, robotics and lasers is the only solution that enables farmers to reduce their most expensive line item – weed control – without damaging plants or the soil,” said Rob Freelen, managing director of Sozo Ventures. “I am particularly impressed with the team’s fast pace of innovation to bring breakthrough products to market, boosting farmers’ profitability across conventional, organic and no-till practices.”

The LaserWeeder identifies weeds and targets them for elimination. The implement’s 30 high-powered CO2 lasers use thermal energy to destroy the meristem of the weed with millimeter accuracy, without damaging nearby crops or disturbing the soil.

“This financing round further supports our mission to provide cost-effective and efficient precision ag-tech tools to growers,” said Paul Mikesell, CEO and founder of Carbon Robotics. “Traditional weeding methods, including hand weeding and herbicides, are expensive, unreliable and damage soil health. The LaserWeeder uniquely addresses all of these challenges.”

To date, Carbon Robotics’ LaserWeeders have successfully eliminated more than 500 million weeds across 40 different crops. This year, the LaserWeeder will be delivered to farms across 17 U.S. states and three provinces in Canada. Most recently, Carbon Robotics expanded the features of the LaserWeeder with the industry’s first LaserThinning capability, targeting areas where vegetable crops are purposefully overseeded and then thinned for optimal crop spacing, growth and yield.

As part of the Series C financing, Rob Freelen with Sozo Ventures and Erik Benson with Voyager Capital will join the Carbon Robotics board of directors.

The post Carbon Robotics raises $30 million in series C funding to scale LaserWeeder platform appeared first on The Robot Report.

Amazon has released a dataset that contains images of more than 190,000 objects that it said can be used to train robots for pick-and-place tasks. Amazon claims this is the largest dataset of images captured in an industrial product-sorting setting.

The dataset, called ARMBench, can be used to train pick-and-place robots to better generalize new objects and contexts. The images were collected in an Amazon warehouse where a robotic arm retrieves a single item from a bin full of items and then transfers it to a tray on a conveyor belt. This task can be difficult because of the variety of objects in the bin and their various configurations and interactions.

Images in the dataset fall into three categories:

• Pick images: top-down images of a bin filled with items before a robot starts picking
• Transfer images: images captured from multiple viewpoints as the robot transfers an item to the tray
• Place images: top-down images of the tray in which the selected item is placed

ARMBench contains images from three separate tasks, object segmentation, object identification and defect detection.

The object detection dataset, which helps robots identify the boundaries of different products in the same bin, contains more than 50,000 images. The images show anywhere from one to 50 manual object segmentations per image, with an average of about 10.5.

The object segmentation dataset helps robots determine which product image in a reference database corresponds to the highlighted product in an image. This dataset includes more than 235,000 labeled pick activities, with each pick activity including a pick image and three transfer images. This dataset also includes reference images and text descriptions of more than 190,000 products. Models can learn to match one of these reference products to an object highlighted in pick and transfer images.

From left to right: a pick image, a transfer image and place image from Amazon’s ARMBench dataset. | Source: Amazon

The defect detection dataset, which includes both images and videos, helps systems know when a robot has committed an error, like picking up multiple items rather than one or damaging an item during transfer. The dataset has more than 19,000 images captured during the transfer phase. It also includes more than 4,000 videos that document pick-and-place activities that resulted in damage to a product.

Videos are a key aspect of this dataset, as certain types of product damage are best diagnosed through video, as they can occur at any point in the transfer process. The defect detection dataset also contains images and videos for over 100,000 pick-and-place activities without defects.

Amazon plans to continue to expand the number of images and videos, and the range of products they depict, in ARMBench.

In November 2022, Amazon unveiled Sparrow, a robotic arm capable of picking individual products before they get packaged. Sparrow can pick 65% of the over 100 million different items that could be processed at an Amazon warehouse, according to the company.

Sparrow can pick a variety of items, like DVDs, socks and stuffed animals, but struggled with items that have loose or complex packaging. It seems likely the company drew on the research it did while developing Sparrow to build this dataset.

The post Amazon’s ARMBench dataset helps train pick-and-place robots appeared first on The Robot Report.

Covariant raised another $75 million in its Series C funding round. The company announced the first part of its series C round, totaling $80 million, in 2021. It has now raised $222 million since its founding.

Radical Ventures and Index Ventures, both returning investors in the company, co-led the round. The most recent round also included participation from other returning investors Canada Pension Plan Investment Board and Amplify partners, and new investors Gates Frontier Holdings, AIX Ventures and Northgate Capital.

Since its founding in 2017, Covariant has focused on developing the Covariant Brain, which the company called a Universal AI platform. Pre-trained on millions of picks from Covariant robots in warehouses around the world, the Covariant Brain enables robots to autonomously pick many SKUs. Covariant plans to use the funding to ensure its platform will allow retail and logistics providers to deploy robotic picking quickly.

“The leading companies have turned to AI Robotics to automate their most manual operations in order to decrease labor costs, increase throughput, and control profitability,” said Covariant CEO Peter Chen. “The past year for Covariant has been incredible with 6x growth in 2022 – and we are just getting started. This infusion of new capital allows us to scale even faster, ensuring more retailers can automate more parts of their fulfillment networks to remove manual bottlenecks, handle fluctuating demand, and better prepare for ever-changing business needs.”

Since its last funding announcement in 2021, the company has applied Covaraint Brian to a broad set of piece-picking and case-picking applications, including order sortation, item induction, good-to-person order picking, knitting and depalletization. With the platform, connected robots learn as a fleet, enabling operational improvements to automatically propagate across customers’ networks.

“Many companies are trying to break into the AI robotics space, but Covariant has been making significant progress for years now,” said Mike Volpi, partner at Index Ventures. “I’m confident that their team, which represents the best minds in AI, and their approach of deploying a unified AI platform are shaping the future of automation, and look forward to the additional progress they’ll make in the years ahead.”

The post Covariant raises $75M for robotic picking technology appeared first on The Robot Report.

A Cruise robotaxi failed to brake quickly enough behind a city bus in San Francisco. | Source: Cruise

Cruise issued a voluntary recall with the National Highway Transporation Safety Administration (NHTSA) in response to a minor collision where a Cruise autonomous vehicle (AV) hit the back of a San Francisco bus. The recall affects 300 AVs. 

The Cruise AV involved in the crash failed to brake quickly enough after the city bus in front of it slowed, according to the company. While the vehicle did brake, it applied the brakes too late and rear-ended the bus at about 10 MPH, Cruise said.

After the collision, Cruise began an investigation and found the cause of the crash was an error related to predicting the movement of articulated vehicles, which are vehicles with two sections connected by a flexible joint allowing them to bend in the middle, like the bus in the accident. 

According to Cruise, the robotaxi saw the front section of the bus and recognized that it was an articulated vehicle that could bend, so it predicted the bus would move as connected sections with the rear section following the predicted path of the front section. As the bus pulled out, Cruise said the AV reacted based on the predicted actions of the front end of the bus, which it could no longer see, rather than the actual actions of the rear section of the bus, making it slow to brake. 

Once the company found the root cause of the accident, it started working on a software update that it said would improve performance near articulated vehicles. When the update was completed, tested and validated, Cruise’s operations team rolled the change out to the fleet, just two days after the incident occurred. Results from testing indicate the specific issue that caused the accident won’t recur after the update. 

“Our data and simulations showed that it was exceptionally rare. At the time of the incident, our AVs had driven over 1 million miles in fully driverless mode. We had no other collisions related to this issue, and extensive simulation showed that similar incidents were extremely unlikely to occur at all, even under very similar conditions,” Cruise founder and CEO Kyle Vogt wrote in a blog. “The collision occurred due to a unique combination of specific parameters such as the specific position of the vehicles when the AV approached the bus (with both sections of the bus visible initially, and then only one section), the AV’s speed, and the timing of the bus’s deceleration (within only a few seconds of the front section becoming occluded).

“We will undoubtedly continue to discover ways in which we can improve, even if that involves changing software that is currently deployed in the field,” he continued. “We think any potential improvement to roadway safety is worthwhile, and we will approach it with the same level of rigor as we’ve demonstrated here. These continuous improvements are likely to make voluntary recalls commonplace. We believe this is one of the great benefits of autonomous vehicles compared to human drivers; our entire fleet of AVs is able to rapidly improve, and we are able to carefully monitor that progress over time.” 

The post Cruise recalls 300 robotaxis in response to crash with bus appeared first on The Robot Report.

Gatik recently announced a multi-year commercial collaboration with Kroger to transport customer orders within Kroger’s Dallas distribution network. 

Starting sometime in Q2 of 2023, Gatik’s medium-duty autonomous box trucks will be transporting products from a Kroger Customer Fulfillment Center (CFC) in Dallas, Texas to multiple retail locations. Gatik’s autonomous trucks will operate with safety drivers in the vehicles for now, but the company plans to eventually take them out. 

Gatik’s trucks feature a cold chain-capable 20′ foot box to transport ambient, refrigerated, frozen goods. The collaboration involves consistent, repeated delivery runs multiple times per day, seven days per week, across Kroger’s Dallas distribution network. Each trip, from a CFC to a retail location, typically involves around 60 miles of driving round trip, according to Gatik’s Head of Policy and Communications Richard Steiner. 

The company has always been interested in automating the middle mile. Steiner joined the company in 2019, just two years after it was founded, and has helped it grow into a now 150-person team. 

“What we’ve seen over the last few years, even prior to the pandemic, is e-commerce going through the roof,” Steiner said. “Consumers like you and me no longer want to wait three, four, or five days for goods, we want them within a one to two-hour pickup window.” 

According to Steiner, this change in the way people shop has led to smaller distribution centers, more micro fulfillment centers and more customer fulfillment centers closer to where those customers live. This is where automating the middle mile, between these fulfillment centers and retail locations, can greatly increase the speed and number of orders fulfilled. 

“A really, really critical point to note about what we’re doing there is we’re increasing the delivery frequency,” Steiner said. “So that means that Kroger’s customers have a greater range of same-day pickup times as well as greater flexibility with cut-off times to place their orders.”

Automating the middle mile also means getting autonomy into the hands of Gatik’s customers quickly, because the team only needs to operate its AVs on repeatable routes instead of anywhere in a city a person might want to go. 

“Compared to the other applications of technology, either long haul or passenger transportation, [middle-mile transportation] is simpler,” Steiner said. “We are constraining the challenge of autonomy by focusing on fixed known repeatable point-to-point routes. So whereas the passenger transportation models require mapping out an exponentially larger and larger geo-fenced area to serve an increasing number of consumers . . . We focus on a limited number of pickup locations and drop off locations, which means that we know our routes more intimately than anyone else on the planet.”

While Gatik doesn’t have a specific timeline for when it will be able to pull out its safety drivers from its trucks in Texas, with each deployment, including its deployment with Walmart in Arkansas and with Loblaw in Ontario, the process keeps moving faster.

“With Walmart in Arkansas, we began commercial operations in June of 2019. We pulled the driver out in August of 2021, so about two years right?” Steiner said. “With Loblaw, we began commercial operations in January 2020 and pulled the safety driver out 19 months after that. So we went from 24, 25 months to 19.”

Gatik has three phases that it works through for deployments. The first involves millions of miles of simulated training, which is followed by the second phase, which involves private closed course track testing, first with safety drivers and then without. Finally, the Gatik team moves into its public testing phase. 

With or without a safety driver, Gatik’s trucks begin generating revenue for its customers on day one. 

“We’ve now done over half a million customers orders across our customer base in North America,” Steiner said. “Every single one of those deliveries has been revenue generating. So we’re not testing for testing’s sake. This is a real business. This is a revenue-generating business.” 

The post Gatik to start deliveries with Kroger in Texas appeared first on The Robot Report.

Yaskawa’s Pallet Builder extension (left) and a Yaskawa robot using the company’s new MotoPick 4 software (right). | Source: Yaskawa

Yaskawa released its MotoPick 4 software system that adds no-code capabilities for high-speed pick and place. The company is also expanding its Smart Pendant’s capability with its new Pallet Builder extension. 

With the building blocks to create systems for up to 16 robots and up to 6 conveyors or 16 tables per robot, the Yaskawa MotoPick 4 software can synchronize multiple robots equipped with vision to pick fast-moving products off a conveyor and place them on an outfeed device, tray or box in an organized arrangement.
 
MotoPick 4 provides precise control and coordination for conveyor speed, accommodating tracking speeds up to 1 M/sec. If the advancement of either the infeed or outfeed conveyor falls behind, the software can be configured with an area that slows or stops conveyors until the operation is equalized. Additionally, the software features a pause function that allows for the system to be paused and later resumed without losing track of item positions.
 
Dynamic load balancing for single or multiple robot systems is accommodated. The ability to asynchronously pick up to four parts with multiple placements per cycle is also offered, with up to 32 different product types to be easily picked and sorted. The gripper is allowed up to eight zones.
 
Supporting multiple encoders and cameras, a robot interface board reduces the integration wiring and panel real estate required for multiple robot cells. Cameras are supported based on conveyor width, workpiece size, target accuracy, production volume and conveyor speed, and they do not need to be aligned in a pre-defined way.
 
Multi-layer tray support allows multiple-layer pattern picking or placing with trays. MotoPick 4 is compatible with the FS100 and YRC1000micro controllers, as well as the MLX300 controller software solution.

Pallet Builder

Expanding the Smart Pendant capability, Pallet Builder enables the development and deployment of palletizing cells. Designed for single pick-and-place tasks, this user interface uses a robust database and guided prompts to facilitate no-code programming for work cell configuration and pallet pattern setup.
 
Designed for small- to medium-sized enterprises with high-mix, low-to-medium-volume production that may have insufficient access to experienced robot programmers, Pallet Builder provides general functions for cell definition, gripper definition, and single pick and place jobs. Multiple infeeds, build stations and dispenser stations (for slip sheet separators) support up to eight combined stations.
 
Job editing is available for specific system requirements. Pallet Builder can accommodate a variety of grippers and end-of-arm tooling and adjusts for product variation. Capable software includes multiple automatic pattern styles as well as full custom pattern support. Single drop positions with selectable label orientation are also available.
 
Compatible with the YRC1000 and YRC1000 micro controllers, Pattern Builder is available as a complimentary download for use with GP-series and HC-series robots. Smart Pendant V3.0 or newer is required.

Yaskawa Motoman launched its newest robot in the PL-series line, the PL800, in October 2022. This industrial robot features an 800 kg (2,140 lb) payload capacity and is designed for a variety of palletizing applications, layer picking, and other logistical tasks for end-of-line or distribution automation.

The post Yaskawa releases MotoPick 4 software & robotic pallet builder appeared first on The Robot Report.

Geek+ P-series robots handle storage shelving then queue and present items for picking to humans. | Credit: Geek+

Geek+ recently announced its combined worldwide fleet of P-series goods-to-person (GTP) workflow robots picked ten billion items over the past year. The GTP picking solutions deployed around the world covered over 175 million km during the period, according to the company, which is further than the distance from the Earth to the sun.

Yong Zheng, Founder and CEO of Geek+, said: “We are very proud of what we have achieved recently. The amount of merchandise that our robots have handled, coupled with the savings in time and energy, demonstrate that mobile robots are a technology for today’s problems and a brighter future.”

Each day, the number of items managed by Geek+ GTP robots and warehouse management systems reached as much as 750 million, according to the company.

This is a significant milestone for a fleet of robots. In researching this claim, we reached out to the company to inquire about the deployed number of GTP robots used to reach this milestone.

In the U.S., Geek+ said it has deployed more than 4,000 robots in 24 locations since 2019. Worldwide, the company said it has deployed an additional 26,000 robots for a total worldwide fleet of 30,000 robots. We used that final figure in the calculation of this milestone.

With all of the data in hand, each robot needed to pick an average of 1.5 items per minute and drive at an average speed of 440 mm/sec (well within the technical and operating specifications of the robots).

The table below contains output from the spreadsheet that we generated to validate these claims. A company representative stated that the formula for calculating the items picked number was calculated using the average tasks picked per hour * average pieces per task * 10 working hours per day * operating robot * operating period.

Table 1 – Calculations for Geek+ robots (initial conditions in bold)

*data provided by Geek+

Over the past year, Geek+ has added new projects and expanded its collaboration with existing clients while continuing to evolve the complete solution. The company optimized Geek+’s warehouse management systems to boost robot efficiency by 15%. Geek+ said it plans to continue to improve and enhance its technology and expects further gains in sustainability and efficiency in 2023.

In 2022, Geek+ secured partnerships to expand into Canada and Brazil. Also, Geek+ has expanded its operations by adding a 1400m2 experience center and two new offices in North America, with over 100 employees.

The post How Geek+ robots could’ve picked 10B items in a year appeared first on The Robot Report.

On the left, AutoStore’s ASRS, and on the right, Ocado’s ASRS. | Source: AutoStore, Ocado

Yesterday, the UK High Court ruled that AutoStore’s patents, which have been the subject of a patent infringement case against Ocado, were invalid, and that Ocado did not infringe upon them. Both AutoStore and Ocado operate automated storage and retrieval systems (ASRS) that operate on similar principles.

AutoStore filed the lawsuit in October 2020, when it asserted that Ocado infringed upon six of its patents. The European Patent Office invalidated two of these patents before judgment was made on the case, and two other patents were withdrawn by AutoStore before the hearing started. 

Judge Hacon invalidated the remaining two patents because AutoStore had disclosed its technology publically in a business deal with the Central Bank of Russia before the company filed its patents. Halcon ruled that even if the patents were valid, the Ocado Smart Platform (OSP) did not infringe on them. The judge also ruled that OSP didn’t infringe upon the patents that AutoStore chose to withdraw from the case. 

An Ocado spokesperson said, “As we have said consistently since the day the action was launched that we did not infringe any valid AutoStore patents. Once again a judge has proved we are right. And we have been proved right by courts both in the US and the UK. Autostore’s decision to sue us has been a complete waste of time – for us and them – and will now also be a further waste of money for them as we intend to seek a significant costs order against them.”

“This entire misguided exercise by AutoStore has simply served to show that it is Ocado that is the innovation factory with robust processes in place to protect and respect IP. Our fulfillment centers can freely operate in light of this judgment. And although AutoStore presented this litigation to the market as a one-way bet – either they would win or “the status quo would be maintained”, the actual outcome of the litigation is that a significant number of their patent assets are being invalidated, their patent portfolio has been reduced and their IP has been weakened. Actions continue against AutoStore in Germany and the US for infringement of Ocado IP rights. The risk of infringement is now with AutoStore.”

In December 2021, AutoStore lost a similar patent infringement lawsuit filed with the International Trade Commission (ITC). US ITC Judge Charles Bullock ruled three of AutoStore’s patents involved in the case were invalid, and that Ocado didn’t infringe upon the fourth patent in the trial. 

The ASRS systems from both companies operate on similar storage and retrieval principles. Items are stored in bins stacked in a grid and retrieved by mobile robots from above the stack.

The post Ocado wins UK patent lawsuit over AutoStore appeared first on The Robot Report.

Bill Gates recently had the opportunity to take a ride in London inside a Wayve autonomous vehicle (AV) and gave his thoughts on the future of driving in a new Gates Notes blog post.

While most AVs work by extensively mapping out the areas they drive in, Wayve uses deep learning techniques to allow its cars to drive anywhere humans can. To train its software, Wayve uses hundreds of millions of data samples of real-world and simulated driving. 

Last year, Wayve announced it was working with Microsoft, which Gates co-founded, to leverage the supercomputing infrastructure needed to support the development of AI-based models for AVs on a global scale. Microsoft also participated in the company’s $200 million funding round in January 2022. 

The Wayve AV was able to safely navigate the busy streets of downtown London with a safety driver present behind the wheel. 

In his blog post, Gates seemed cautiously optimistic about the future of AVs. He predicted that it’s likely the AV industry will reach a tipping point between SAE level 2 and 3 vehicles, or, in other words, vehicles that can only provide assistance to drivers through autonomous features, and vehicles that take the control out of the drivers’ hands. 

“Over the next decade, we’ll start to see more vehicles crossing this threshold. AVs are rapidly reaching the point where almost all of the technology required has been invented. Now, the focus is on refining algorithms and perfecting the engineering,” Gates wrote in his post. “There have been huge advances in recent years—especially in sensors, which scan the surrounding environment and tell the vehicle about things it needs to react to, like pedestrians crossing the street or another driver who swerves into your lane.”

Gates predicts that the first AVs that will see widespread adoption will be for long-haul trucking, then deliveries, and then autonomous taxis and rental cars will become common.

Widespread adoption of AVs, however, will come with other issues. For example, when an AV is involved in an accident, insurance companies will have to determine if the blame falls on the passenger or the company that programmed the car software. Gates predicts that these kinds of changes, along with regulatory changes, could take decades to fully come to fruition. 

The post Watch Bill Gates take a ride in a Wayve AV appeared first on The Robot Report.

What is the best architectural approach to use when developing medical robots? Is it ROS, ROS 2 or other open-source or commercial frameworks? The upcoming Robotics Summit & Expo (May 10-11 in Boston) will explore engineering questions concerning the level of concern, risk, design controls, and evidence on a couple of different applications of these frameworks.

In a session on May 10 from 2-2:45 PM, Tom Amlicke, Software Systems Engineer, MedAcuity will discuss the “Keys to Using ROS 2 and Other Frameworks for Medical Robots.” Amlicke will look at three hypothetical robotic systems and explore these approaches:

• 1. An application based on the da Vinci Research Kit through regulatory clearance
• 2. ROS as test tools to verify the software requirements for a visual guidance system
• 3. Commercial off-the-shelve robot arm used for a medical application

If you attend his session, you will also learn how to create trade-offs with these different architectural approaches and how to validate the intended uses of these architectural approaches to ensure a successful submission package for your FDA, EMA, or other regulatory approval.

Amlicke has 20-plus years of embedded and application-level development experience. He designs and deploys enterprise, embedded, and mobile solutions on Windows, Mac, iOS, and Linux/UNIX platforms using a variety of languages including C++. Amlicke takes a lead role on complex robotics projects, overseeing end-to-end development of ROS-based mobile robots and surgical robots.

You can find the full agenda for the Robotics Summit here. The Robotics Summit & Expo is the premier event for commercial robotics developers. There will be nearly 70 industry-leading speakers sharing their development expertise on stage during the conference, with 150-plus exhibitors on the showfloor showcasing their latest enabling technologies, products and services that help develop commercial robots. There also will be a career fair, networking opportunities and more. 

The post Keys to using ROS 2 & other frameworks for medical robots appeared first on The Robot Report.

Gecko Robotics, a creator of wall-climbing robots for asset inspection, announced that it’s expanding its work with the US Navy. Gecko’s robots will now be inspecting the US Navy’s first amphibious assault ship and an additional Arleigh Burke Class destroyer. 

Gecko’s AI-powered robots gather data and build digital models of the Navy’s vessels to increase the speed of maintenance cycles and reduce the amount of time Navy vessels spend in dry dock. 

Since being deployed with the Navy, Gecko’s robots have been able to reduce the lead time required for a ship rudder inspection from 11 days to one. Gecko’s inspections are also much more thorough than manual ones, for example, for one Navy asset, Gecko’s robots found more than 4.2 million data points on its inspection, while traditional methods found only 100 data points. 

“We built Gecko Robotics to solve the hardest physical problems facing the world’s most important organizations,” Jake Loosararian, CEO and Co-Founder of Gecko Robotics, said. “We are proud to have a mature technology that has been tested and approved by both Navy technical leaders and the sustainment officials charged with reducing the Navy’s maintenance backlog. The sailors of the U.S. Navy have a vital mission in an increasingly complicated geopolitical environment and Gecko stands with them to make sure they have the tools they need to do their jobs safely and effectively.”

Gecko isn’t just working with the US Navy, in December 2022, the company announced it was awarded an 18-month, $1.5 million contract through the US Air Force Nuclear Weapons Center’s Small Business Innovation Research (SBIR) program.

Gecko Robotics’ technology is made for the power generation, oil and gas, heavy manufacturing and defense industries. Its TOKA Series robots, which include the TOKA 3, TOKA 4, TOKA 4 GZ and TOKA Flex, are each designed to perform under certain circumstances.

The TOKA 3, for example, is better for inspections on medium-sized piping and high-temperature surfaces, while the TOKA 4 does better with boiler walls and curved surfaces. The TOKA 3 is able to cover 60 feet per minute, while the others in the series can reach a top speed of 30 feet per minute.

Gecko was co-founded by Loosararian and Troy Demmer, now the chief product officer. Loosararian started the company in 2013 and in 2016 joined efforts with Demmer. The company received $2.1 million in funding the same year. 

The post Gecko Robotics expands work with U.S. Navy appeared first on The Robot Report.

Amazon recently detailed how Astro, the company’s multi-purpose home robot, can navigate through its environment with limited onboard computational capabilities. Astro’s sensor field of view and onboard computational capabilities aren’t nearly as powerful as other autonomous robots. While this makes it a more affordable option for consumers, it also means it’s more challenging for Amazon to deliver a high-quality of motion. 

Amazon counteracts Astro’s lack of computation capabilities with algorithms and software designed to allow the robot to move more gracefully. 

Predictive planning is a key aspect of Astro’s navigational abilities. Astro’s limited computational capabilities mean it struggles with a large sensing-to-actuation latency. To combat this, Astro makes predictions about the movements of the objects around it, like people. The robot predicts where those objects will be and what its surroundings will look like at the end of its current planning cycle, helping it to account for latencies in sensing and mapping while it’s moving.

All of Astro’s plans are based on its latest sensor data and what it thinks its surroundings will look like when its plan will be taking effect. The robot can make these predictions because of its ability to predict and handle uncertainties and risks of collisions. 

Astro’s motivation to move towards its goal is always weighed dynamically with its perceived level of uncertainty. This means Astro evaluates uncertainty-adjusted progress for each candidate motion, allowing it to focus on getting to its goal when it determines risk is low, and focus on evasion when risk is high. 

The robot also uses trajectory optimization software to operate in its environment. Astro considers multiple candidate trajectories and picks the best one in each planning cycle. The robot plans 10 times a second and evaluates a few hundred trajectory candidates in each instance. 

Astro considers safety, smoothness of motion and progress toward its end goal. With these three criteria, the robot picks the trajectory that will result in optimal behavior. Other approaches limit the number of choices a robot can make to a discrete set, or a state lattice, but Amazon’s formulation is continuous, helping the robot move smoothly. 

Astro doesn’t just have to plan where its two wheels and body will go, it also has to plan movements for Astro’s screen. The robot’s screen is used to communicate motion and intent and for active perception, so Astro plans to do things like orienting its screen towards the person it’s following or in the direction it plans to go so humans around it know what its plans are. 

Amazon released Astro in September 2021. The robot can be used for a variety of things, including home monitoring, videoconferencing with family and friends, entertaining children, and more. The voice-controllable robot can recognize faces, deliver items to specific people, after a human puts the item in the storage bin, and use third-party accessories to, for example, record blood pressure. It can detect the sound of a smoke alarm, carbon monoxide detector or breaking glass. If you have a Ring account, Astro can send you notifications if it notices something unusual.

The post How Amazon Astro moves through its environment appeared first on The Robot Report.

Last Thursday, March 23, 2023, a Cruise robotaxi, operating autonomously and with no driver or passengers in the vehicle, was involved in a minor accident with a San Francisco Muni bus. No one was injured in the incident. Both the Cruise vehicle and Muni bus received some damages, with the robotaxi seeming to take the brunt of the impact, according to Forbes.

While there aren’t many details about the accident, Cruise tweeted the following brief statement after the incident: “Yesterday, one of our vehicles made contact with the rear of a Muni bus. No injuries were reported, and there were no passengers present in the Cruise AV. We are investigating the incident and will take follow-up actions if appropriate.”

Robotaxi companies are typically quick to clear things up if their vehicles aren’t at fault in incidents. So many are assuming the Cruise robotaxi rear-ended the bus, based partly on Cruise saying in the statement one of its vehicles “made contact with the rear of a Muni bus.”

This incident happened just days after another Cruise robotaxi drove through caution tape that closed off a road where overhead trolley wires had fallen during a major high-wind storm. The vehicle hit the wires which were powered off at the time. Had the power lines been live when the vehicle hit them, the incident could have been much worse.

In December, the National Highway Traffic Safety Administration (NHTSA) announced it was opening an investigation into the company’s automated driving system. In a filing, NHSTA said it was interested in two different issues that had been reported to the administration that both result in the robotaxis becoming hazards for others on the road. 

NHSTA’s first concern is with the company’s robotaxis reportedly braking too hard in certain circumstances. The administration received three reports of a Cruise vehicle braking too hard in response to a car approaching them quickly from the rear, resulting in the Cruise vehicle being rear-ended in all three incidents. 

The administration’s second concern is with the robotaxis becoming immobilized on the road. NHSTA has received several reports of Cruise vehicles stalling on the road, with no human operators present. NHTSA’s Office of Defects Investigation (ODI) opened a preliminary investigation to determine the scope and severity of both issues. 

Last week, Cruise filed an application with the California DMV to test its robotaxis statewide. The company has been working to expand its services in several areas, including in Austin and in the suburbs of Phoenix. It’s also working to roll out its purpose-built Origin vehicle on public roads in San Francisco. 

Last month, Cruise announced its fully driverless vehicles have traveled over one million miles. A majority of Cruise’s fully driverless miles were driven in San Francisco, where the company first rolled out its robotaxis in November 2021. 

The Robot Report reached out to Cruise for details on both incidents but hasn’t heard back as of press time. If we do get any more information, this story will be updated.

The post Cruise robotaxi, SF bus involved in accident appeared first on The Robot Report.

The first 3D-printed rocket called GLHF from Relativity Space was successfully launched this week. | Credit: Relativity Space

Relativity Space successfully launched a 3D-printed rocket on Wednesday March 24, 2023. The first Terran 1 rocket called “GLHF” (Good Luck Have Fun) lifted off from launch complex 16 at Cape Canaveral. The Terran 1 is the largest 3D-printed object to be launched as a rocket.

After two prior scrubbed launch attempts over the last week, GLHF successfully lifted off from the launch pad, and reached two goals during its brief flight:

1. Max-Q: the point of greatest aerodynamic pressure on the rocket body
2. Main engine shut off: the completion of the main engine burn

The rocket failed to reach orbit in what appears to be a secondary rocket engine failure. As of press time, there is no news on the exact failure. Without the secondary engine ignition, the rocket did not have enough power to reach orbit.

Relativity Space uses industrial robots to 3D print rocket’s body

Relativity Space is innovating through the use of robotics to 3D print metal items. As shown in the video below, the company uses industrial robot arms to manage the application of welded metal to produce the rocket body and rocket motors. Rocket bodies are currently expensive to manufacture, and 3D printing them promises to reduce costs and reduce weight by creating complex shapes that require less material (by mass) than with current manufacturing methods.

Traditional methods might take 24 months to produce a rocket body, whereas 3D printing can reduce that timeframe to 2 months. Likewise, a 48-month iteration time on a complete rocket assembly can be reduced to 6 months.

The Terran 1 is designed for the future of constellation deployment and resupply. With software-driven design and the use of 3D printing, the payload part of the rocket (in the nose cone) can easily be manufactured to the specs of each individual launch client.

Ultimately, Relativity Space is planning to produce the Terran R reusable robot, further improving the ROI on rocket development as well as the cost per launch. The vision is to create a fleet of reusable rockets to support future missions to the moon, mars and beyond.

The company was founded in 2015 and has raised $1.3B to date. (Crunchbase)

Here’s the live stream of the Terran 1 rocket launch:

The post Relativity Space launches first 3D printed rocket appeared first on The Robot Report.

Here’s a recap, in no particular order, of some of the best robots at ProMat 2023. 

Agility Robotics’ Digit

The Agility Robotics booth was packed for most of the show, and for good reason. It’s not every day you get the opportunity to see a bipedal robot in action. According to Agility, Digit’s demo was based on a real customer use case, and it was running during all four days of the show. Digit was the winner of ProMat’s Best New Innovation Award.

Digit is the product of over two decades of research and development. While the company may have started its journey by looking at various use cases for a bipedal robot, anything from running into burning buildings to last-mile delivery, it honed its focus on automating a process that needs to be addressed that Digit can safely and reliably perform. 

Digit currently only works in enclosed spaces away from humans, and it will even shut down when human workers get too close. The robot can help reduce worker injuries from lifting or lacerations that can happen when human workers handle broken totes. 

Agility is gearing up to start manufacturing Digit at scale, adding it could work at a rate of $10/hour for tasks like the one it demoed at ProMat. 

Boston Dynamics’ Stretch 

Boston Dynamics showed off its Stretch case handling robot at the show. Stretch showed off its trailer unloading abilities flawlessly, although there wasn’t much variability in the boxes it was working with. 

Stretch is already deployed at customer sites and has more customers lined up. Boston Dynamics seems to have even more ambitious plans for the robot in the future. At its booth, it ran a video where it showed animated Stretch robots zipping around a warehouse, building pallets on top of AMRs which bring those pallets to other Stretch robots that would load the boxes into a truck. 

The company is being held back by safety regulations right now, which keep Stretch caged off and away from humans, limiting it to just its current unloading application.

Pickle Robot demoed a similar container unloading system. It is built around a modified Kuka arm and suction grippers to pick up objects up to 65 pounds and place them onto a conveyor. The onboard vision system determines which box to pick next. Again, there wasn’t much variety in the shape and size of the boxes. Here’s a side-by-side comparison of the Pickle and Boston Dynamics solutions:

RightHand Robotics’s RightPick

RightHand Robotics wasn’t the only company showing off bin-picking robots, but its demo was one of the most consistent and flexible. The entire time I was at its booth, and every time I visited it throughout the show, RightPick didn’t miss a single pick. 

RightHand also showed a new autonomous suction cup swapper. Integrated in the front of a demo workcell, the robot arm can choose from four different suction cups. Due to all the data RightHand has collected over the years, it knows which suction cup offers the best chance at successfully picking an item. When the system has low confidence in a suction cup’s ability to grasp upcoming items, it will autonomously swap out for the suction cup that offers the most confidence. RightPick’s grippers use a combination of suction and three mechanical fingers to give the system extra stability in all of its picks. 

Brightpick’s Autopicker

Autopicker was announced in February 2023, and it was exciting to see it at work in person for the first time. Brightpick’s Autopicker moves around warehouses, retrieving totes from shelving and robotically picking items from those totes to consolidate orders directly in aisles. This process removes the need for a human picker to take items out of totes, a common part of many similar robotic systems.

At ProMat, Autopicker was picking snacks for attendees. At the show, it was only pulling totes from one side of the robot. But the version that will work in customer warehouses can pull from both sides, allowing it to roll between shelving and switch between pulling totes from behind or in front of the robot. 

Mujin’s TruckBot

Truck unloading was a popular demo at ProMat, and Mujin took a more unique approach to the process. According to the company, the robot can unload up to 1,000 cases per hour.

TruckBot’s design aims to specifically address the challenges that come specifically with unloading floor-loaded trailers and containers. The robot can attach to standard telescoping conveyors, which are found at many loading docks, and can reach up to 52 feet into a trailer truck or shipping container.

At ProMat, TruckBot was working with Mujin’s palletizing robot to sort cases onto pallets for storage. 

Tuskrobots’ autonomous pallet handler

Autonomous forklifts were another common appearance at the show, but Tuskrobots had the most unique approach to moving pallets. Instead of the typical automated forklift design, Tuskrobots’ autonomous pallet handler slides two prongs underneath a pallet and uses the tongs to lift the pallet. The rest of the AMR then slides underneath the pallet. 

Once the pallet is on top of the AMR, it can drive to its next location, where it lifts it again, slides out from underneath the pallet, lowers the pallet, and then slides its prongs back into the AMR. This unique method of picking up pallets means the robot is only slightly larger than its payload, allowing it to operate in high-density storage areas. 

Verve Motion’s SafeLift Suit

Verve Motion spun out of Harvard in 2020 and offers an exosuit that aims to take 40% of the strain off of workers during lifting. While the exosuit can improve productivity by helping workers keep a consistent pace during their entire shift, the company’s aim is to make workplaces safer for workers and to reduce on-the-job injuries. 

When we were at the booth, Home Depot was demoing the system, repeatedly lifting a packed tote and case of water. The retailer said they have employees who repeatedly lift loads up to 75 lbs from the floor onto a pallet daily. 

I had the opportunity to try out the exosuit myself. It took just minutes to put on and was easy to adjust. When picking up and putting down a heavy tote, it gives a tug across the back that reduces the amount of strain your back has to do. The suit at ProMat had three settings: neutral setting, lifting only and lifting and putting down. Users can switch between the modes by pressing a button near their shoulder. These settings can be customized according to a customer’s need. 

Slip Robotics AMR

Slip Robotics showcased its automated trailer loading/unloading system (ATLS). The omnidirectional robot is designed to carry up to 8 full pallets and a total of 6 tons. It autonomously drives into a tractor-trailer for transport to the next destination. Three ATLS robots can fit inside a typical tractor-trailer.

Fork truck operators do not need to enter the trailer and instead load and unload pallets onto the Slip ATLS in the open loading dock. The Slip ATLS travels with the load to the destination, where it drives off the trailer, onto the dock for unloading and loading. A loaded set of ATLS robots would then drive onto the trailer for immediate departure. Slip said its robots use a sealed lead acid battery that isn’t subject to federal hazmat restrictions for ground transport.

The post Editor’s Picks: best robots of ProMat 2023 appeared first on The Robot Report.