Many years ago, I worked for a small machine tool company building special-purpose gantry robots for the automotive industry. These were the robot industry's "Wild West" days where crazy ideas were commonplace. Old notions were being challenged, and incredible machines were being built despite the primitive technology available.
Back then, the joke was that the carbon-based equivalent of a robot would be the dumbest creature on the face of the planet. You couldn't just tell a robot to move a stack of lumber or a pile of bricks like you could with a human. You had to teach it the points in space it had to move through and the speed it had to travel to arrive at the first brick. You had to tell it to close its grippers. You had to give it the exact coordinates for the brick's new location, and you had to command it to open its grippers when it got there.
This process had to be repeated for every brick in the pile, and if a new load of bricks was delivered with any one of them out of position, the result could involve angry men in suits writing checks to repair the damage. Even the dullest worker in your company could do better than that, so you had to wonder, what exactly were the circumstances that would cause someone to lose their job to such a brainless device!
Fast forward thirty years, and NVIDIA’s Isaac AMR platform is eliminating the brainless part. The advent of e-commerce combined with persistent worker shortages has created a growing demand for sophisticated robotic supply chain solutions and NVIDIA has stepped in to provide them. The company is predicting that by 2025 there will be 53,000 facilities employing autonomous mobile robots to shuttle material to assembly and shipping areas.
The typical industrial facility is a complex space filled with humans behaving in unpredictable ways. They enter high-traffic areas without looking. They stop and change directions without warning. They block accessways and leave personal equipment in hazardous places. The ability to operate equipment outside human control in this environment is a significant challenge.
NVIDIA's Isaac AMR platform was developed to assist this process by providing mapping, site analytics, and fleet optimization. It starts with the NVIDIA Omniverse, which creates a digital twin of the facility that can simulate the operating conditions for the robot fleet. As part of Omniverse, the Isaac platform can integrate with other components such as Metropolis, ReOpt, and soon DeepMap.
DeepMap is a cloud-based SDK that can dramatically reduce the time required to map a facility while maintaining a high degree of precision. The DeepMap Update Client allows robot maps to be updated whenever necessary in real-time. It also provides robots with a semantic understanding of their current view, allowing them to identify stationary and movable objects.
NVIDIA Metropolis adds real-time situational awareness to your robot fleet. Mapping cannot account for everything in a dynamic environment, and onboard sensors are not always enough to ensure safe and efficient operation. Metropolis provides access to cameras and sensors mounted strategically about the facility. These additional sensors allow AMRs to see around the next corner, eliminating blind spots and congested areas.
NVIDIA ReOpt AI software libraries can be used to optimize vehicle route planning and logistics in real-time, which can be applied to AMR fleets. Robot speeds, battery life, transport size, weight, and facility layout are crucial factors needed to determine the optimal AMR fleet size. ReOpt allows companies to simulate multiple AMR interactions using Isaac Sim before making expensive decisions. After deployment ReOpt can re-optimize existing routes to provide maximum operational efficiency.
The Isaac AMR platform provides a complete solution for integrating AMRs into human environments using advanced industrial and human-robot simulations as well as route optimizations. You can read more about it here.