MIT researchers are looking to bridge the large gap between how quickly robots can process information (relatively slowly) and how fast they can move (very quickly thanks to modern advances in hardware), and they’re using something called “ robotic computing ” to do it. The method, devised by MIT Computer Science and Artificial Intelligence (CSAIL) graduate Dr Sabrina Neuman, results in custom computer chips that can offer hardware acceleration to speed up response times.
Custom chips tailored for a very specific purpose aren’t new – if you’re using a modern iPhone, you have one in this device right now. But they have become more popular as businesses and technologists look to do more local computing on devices with more conservative computing power and constraints, rather than moving data to large data centers over connections. network.
In this case, the process involves creating hyper-specific chips designed according to the physical layout of a robot and its intended use. By taking into account the requirements of a robot in terms of its perception of its environment, its mapping and understanding of its position in these environments, and its motion planning resulting from said mapping and its required actions, researchers can design processing chips that significantly increase the efficiency of this last step by supplementing software algorithms with hardware acceleration.
The classic example of hardware acceleration that most people come across on a regular basis is a graphics processing unit or GPU. A GPU is essentially a processor designed specifically for the task of handling graphics computer operations – such as display rendering and video playback. GPUs are popular because almost all modern computers run in graphics-intensive applications, but custom chips for a range of different functions have become much more popular in recent times thanks to the advent of small-batch chip manufacturing techniques. more customizable and more efficient.
Here is a description of how Neuman’s system works, especially in the case of optimizing a hardware chip design for robot control, from MIT News:
The system creates a custom hardware design to better meet the computing needs of a particular robot. The user enters parameters of a robot, such as the arrangement of its limbs and how its different joints can move. Neuman’s system translates these physical properties into mathematical matrices. These matrices are “sparse,” meaning they contain many zero values that roughly correspond to movements that are impossible given the particular anatomy of a robot. (Likewise, your arm movements are limited because it can only bend at certain joints – it’s not an infinitely pliable spaghetti noodle.)
The system then designs a specialized hardware architecture to perform calculations only on non-zero values in the matrices. The resulting chip design is therefore designed to maximize the efficiency of the robot’s computing needs. And this customization paid off in testing.
Neuman’s team used a Field-Programmable Gate Array (FPGA), which is kind of like a midpoint between a fully custom chip and a standard processor, and it performed significantly better than the latter. This means that if you made a custom chip from scratch, you could expect much larger performance improvements.
Making robots respond to their environment faster isn’t just about increasing manufacturing speed and efficiency – it does. It’s also about making robots even safer in situations where people are working directly alongside and in collaboration with them. This remains a significant barrier to more widespread use of robotics in everyday life, meaning this research could help unlock the sci-fi future of humans and robots living in integrated harmony.