Tech

Combining Unique Gripper Designs With AI and Machine Vision

Righthand Robotics combines machine vision with an intelligent gripper design to provide a more adaptable robot.Credit: Courtesy of researchers

Founded by MIT graduates, RightHand Robotics has developed a more reliable and adaptable picking robot in a warehouse environment.

For most people, it’s easy to identify an object, pick it up, and place it elsewhere. For robots, you need the latest machine intelligence and robot operation.

This is what MIT’s spin-off RightHand Robotics has incorporated into the robot’s piece picking system. The system combines a unique gripper design with artificial intelligence and machine vision to help companies classify products and place orders.

“If you want to buy something in the store, push the cart down the aisle and choose for yourself. Ordering online does the same thing in the fulfillment center,” says Lael Odhner, co-founder of RightHand Robotics. ’04, SM ’06, PhD ’09 states. “Retailers usually need to pick up a single item, scan it, put it in a sorter or conveyor belt to complete an order. There are tens of thousands of orders a day and 10 or 20 football fields. With over 100,000 unique products stored in a sized facility, it sounds easy until you imagine that the estimated delivery time is ticking. “

RightHand Robotics helps businesses respond to two broad trends that have transformed the retail industry. One was the explosive increase in e-commerce, which accelerated only during the Covid-19 pandemic. The other is the shift to just-in-time inventory fulfillment. In this fulfillment, pharmacies, grocery stores and apparel companies replenish their products based on what they purchase that day or week to improve efficiency.

Robotics on the right

“We give people insights into their inventory, insights into how they store their inventory, and how they work both upstream and downstream of the picking we do. It can give you insight into what you’re building, “says Odhner.Credit: Courtesy of researchers

Robot Fleet also collects data to help RightHand Robotics improve the system over time and acquire new skills such as gentler and more accurate placement. Process and performance data is sent to the company’s fleet management software. This allows customers to understand how inventory moves through the warehouse and identify bottlenecks and quality issues.

“The idea is that e-commerce companies can change or overhaul the operational flow of the entire warehouse, rather than just looking at the performance of a single operation,” says Odhner. “The goal is to eliminate upstream fluctuations as much as possible and create a simpler and more streamlined process.”

Push the limits

Odner holds a PhD in the laboratory of Harry Asada, a professor of Ford Engineering in the Department of Mechanical Engineering at MIT. Odner says he encouraged students to become more familiar with robotics research. Colleagues also frequently shared their work at seminars, giving Odner a comprehensive view of the field.

“Asada is a very well-known robotics researcher, and his early work and the projects I worked with him are very basic of what RightHand Robotics is doing,” Odhner said. Says.

In 2009, Odhner was a member of the DARPA Autonomous Robotic and Manipulation Challenge winning team. Many of the competing teams were associated with MIT, and the entire program was finally run by former MIT associate professor Gill Pratt. After the semi-finals of the MIT100K contest as “Manus Robotics” in 2013, the team was introduced to Kiva Systems founder Mick Mountz ’87 (later acquired by Amazon). logistics.

Today, a significant number of RightHand Robotics employees and leadership come from MIT. MIT researchers also explained many early customers by purchasing components invented by Odner’s team during the DARPA program.

“In general, we are so close to MIT that it’s hard to avoid getting back there,” says Odhner. “It’s a kind of family. I never leave MIT.”

At the heart of the RightHand Robotics solution is the idea of ​​using machine vision and intelligent grippers to make peace-picking robots more adaptable. This combination also limits the amount of training required to run the robot, and each machine is equipped with the equivalent of a company’s hand-eye coordination.

“The technical part of what we do is that we need to look at unstructured presentations of consumer goods and make a meaningful understanding of what’s there,” says Odner.

RightHand Robotics also utilizes an arm-end tool that combines suction with a new under-operated finger. This gives Odhner more flexibility than robots that rely solely on suction cups and simple pinch grippers.

“In practice, it can help give the hand a passive degree of freedom, a passive movement that it can do and cannot actively control,” Odhner said of the robot. “Often they simplify control tasks. They remove problems from being overly constrained and make them easier to handle when performing motion planning algorithms.”

The data collected by robots is also used to improve reliability over time and shed light on customers’ warehousing operations.

“We give people insights into their inventory, insights into how they store their inventory, and how they work both upstream and downstream of the picking we do. It can give you insight into what you’re building, “says Odhner. “We have very good insights into what could cause future problems and we can feed them back to our customers.”

Odhner says warehouse fulfillment could grow into a much larger industry if throughput is improved.

“As consumers increasingly value their online shopping options, more and more items need to be added to their’virtual’carts. The availability of people near order fulfillment centers tends to be a limiting factor in e-commerce growth. All of this shows a significant inefficiency in the economy, which is essentially what we are trying to address, “says Odner. “We are performing the least attractive tasks in the warehouse. For example, like sorter guidance, we just pick something, scan it and put it on the belt all day long. We automate these tasks for our customers. Can take your people and you can lead them to what will be felt more directly by the customer. “

Odhner also said that more automated fulfillment centers help protect workers’ health and safety, including ergonomic stations where goods are brought to workers for special tasks and increased social distance. It states that it will provide improved measures. Instead of reducing the number of people working in the warehouse, he says, “Ultimately, I want a system that plays a role such as quality control and supervises robots.”

Robots easy

This year, the company announces a third version of its picking robot. It features standardized integration and safety features to make it easier for warehouse operators to deploy peace-picking robots.

“We don’t necessarily understand the magnitude of our progress in commercializing this autonomous system in terms of ease of integration, configuration, safety and reliability, but we have robotic systems around the world. It’s huge because it means you can dropship to, and you can launch and run it with minimal customization, “says Odhner. “There is no reason why this is in a box or pallet and no one can set it up. That is our big vision.”



Combining Unique Gripper Designs With AI and Machine Vision Source link Combining Unique Gripper Designs With AI and Machine Vision

Related Articles

Back to top button