Every day, FedEx processes about 15 million parcels. Currently, a two-armed robot with AI is expertly packing some of them onto delivery trucks.
This week, FedEx unveiled DexR, a two-armed robot that is intended to automate one of the most challenging jobs facing the company’s human employees—loading a truck with packages.
The new robot intends to increase the number of boxes that can fit within a delivery truck by efficiently stacking rows of variously sized boxes inside the vehicle.
For a machine, such task is not at all simple. According to Rebecca Yeung, vice president of operations and advanced technologies at FedEx, packages come in a variety of sizes, shapes, weights, and packaging materials. The robot must decide how to arrange the available boxes to form a tidy wall, place them securely without crushing anything, and react correctly if any parcels slip after being perceived by cameras and lidar sensors.
“AI wasn’t at the point where it was smart enough to handle this kind of complex decision-making a few years ago,” adds Yeung. DexR is now being tested in preparation for a later, more extensive implementation at FedEx.
While generative AI tools like ChatGPT have given the impression that AI technology is capable of managing almost anything in many industries, addressing items in the complex, unpredictably changing real world still presents significant problems for algorithms. The majority of industrial robots are built to do highly repetitive tasks with unwavering precision.
Roboticists are advancing. AI is now being used by an increasing number of devices to perform tasks like object recognition and grip determination. Before moving the software to a real robot, this may entail training the algorithms in a simulation where errors don’t matter as much. However, transitioning from a simulation to the actual world is notoriously challenging.
Advanced robots are beginning to find more commercial uses because to enhanced algorithms, fresh methods for applying machine learning to robots, and better hardware and sensors.
According to Matthew Johnson-Roberson, head of the robotics institute at Carnegie Mellon University, “in the last year or two, people have taken advancements in AI and machine learning and said ‘we can make a real business case here, whether it’s lowering costs or improving efficiency or whatever.”Robots will begin to appear in more workplaces, according to Johnson-Roberson, who claims that years of investment in fields like self-driving cars, together with a regular rhythm of AI advancements. “I’m hoping that the commercial robotics industry is just getting started.”
Dexterity, a business based in Redwood City, California, that specializes in creating robotic systems for various warehouse jobs using AI, created the FedEx robot for the company.
The system employs force feedback each time it sets a box in a stack to make sure the item fits snugly. It also scans the stack with cameras and depth sensors to compare it to its model. The robot will have to modify its stacking strategy as it goes along if there is any difference.
Working with packages has become a creative area for robot development as a result of the expansion of ecommerce, particularly Amazon. As it works to increase the efficiency of the facilities where it stores and processes products, Amazon is now deploying thousands of increasingly advanced robots.
Fears of job loss may arise if AI speeds up the deployment of robotics. Current technical developments in the car industry, such as electrification and autonomous driving, are partially responsible for the ongoing US autoworker strike.
Yeung claims that although the robot is still being refined, it will eventually be able to load a truck as swiftly as an expert human. Inside some facilities, FedEx already employs robot technology created by another business, Berkshire Grey, to sort packages. on 2022, it invested $200 million on these systems.
