customized hardware, or “brains,” that speeds up a robot’s
operation. Image: Jose-Luis Olivares, MIT
Contemporary robots can move quickly. âThe motors are fast,
and theyâre powerful,â says Sabrina Neuman.
Yet in complex situations, like interactions with people, robots
often donât move quickly. âThe hang up is whatâs going on in
the robotâs head,â she adds.
Perceiving stimuli and calculating a response takes a
âboatload of computation,â which limits reaction time, says
Neuman, who recently graduated with a PhD from the MIT Computer
Science and Artificial Intelligence Laboratory (CSAIL). Neuman has
found a way to fight this mismatch between a robotâs âmindâ
and body. The method, called robomorphic computing, uses a
robotâs physical layout and intended applications to generate a
customized computer chip that minimizes the robotâs response
time.
The advance could fuel a variety of robotics applications,
including, potentially, frontline medical care of contagious
patients. âIt would be fantastic if we could have robots that
could help reduce risk for patients and hospital workers,â says
Neuman.
Neuman will present the research at this Aprilâs International
Conference on Architectural Support for Programming Languages and
Operating Systems. MIT co-authors include graduate student Thomas
Bourgeat and Srini Devadas, the Edwin Sibley Webster Professor of
Electrical Engineering and Neumanâs PhD advisor. Other co-authors
include Brian Plancher, Thierry Tambe, and Vijay Janapa Reddi, all
of Harvard University. Neuman is now a postdoctoral NSF Computing
Innovation Fellow at Harvardâs School of Engineering and Applied
Sciences.
There are three main steps in a robotâs operation, according
to Neuman. The first is perception, which includes gathering data
using sensors or cameras. The second is mapping and localization:
âBased on what theyâve seen, they have to construct a map of
the world around them and then localize themselves within that
map,â says Neuman. The third step is motion planning and control
â in other words, plotting a course of action.
These steps can take time and an awful lot of computing power.
âFor robots to be deployed into the field and safely operate in
dynamic environments around humans, they need to be able to think
and react very quickly,â says Plancher. âCurrent algorithms
cannot be run on current CPU hardware fast enough.â
Neuman adds that researchers have been investigating better
algorithms, but she thinks software improvements alone arenât the
answer. âWhatâs relatively new is the idea that you might also
explore better hardware.â That means moving beyond a
standard-issue CPU processing chip that comprises a robotâs brain
â with the help of hardware acceleration.
Hardware acceleration refers to the use of a specialized
hardware unit to perform certain computing tasks more efficiently.
A commonly used hardware accelerator is the graphics processing
unit (GPU), a chip specialized for parallel processing. These
devices are handy for graphics because their parallel structure
allows them to simultaneously process thousands of pixels. âA GPU
is not the best at everything, but itâs the best at what itâs
built for,â says Neuman. âYou get higher performance for a
particular application.â Most robots are designed with an
intended set of applications and could therefore benefit from
hardware acceleration. Thatâs why Neumanâs team developed
robomorphic computing.
The system creates a customized hardware design to best serve a
particular robotâs computing needs. The user inputs the
parameters of a robot, like its limb layout and how its various
joints can move. Neumanâs system translates these physical
properties into mathematical matrices. These matrices are
âsparse,â meaning they contain many zero values that roughly
correspond to movements that are impossible given a robotâs
particular anatomy. (Similarly, your armâs movements are limited
because it can only bend at certain joints â itâs not an
infinitely pliable spaghetti noodle.)
The system then designs a hardware architecture specialized to
run calculations only on the non-zero values in the matrices. The
resulting chip design is therefore tailored to maximize efficiency
for the robotâs computing needs. And that customization paid off
in testing.
Hardware architecture designed using this method for a
particular application outperformed off-the-shelf CPU and GPU
units. While Neumanâs team didnât fabricate a specialized chip
from scratch, they programmed a customizable field-programmable
gate array (FPGA) chip according to their systemâs suggestions.
Despite operating at a slower clock rate, that chip performed eight
times faster than the CPU and 86 times faster than the GPU.
âI was thrilled with those results,â says Neuman. âEven
though we were hamstrung by the lower clock speed, we made up for
it by just being more efficient.â
Plancher sees widespread potential for robomorphic computing.
âIdeally we can eventually fabricate a custom motion-planning
chip for every robot, allowing them to quickly compute safe and
efficient motions,â he says. âI wouldn’t be surprised if 20
years from now every robot had a handful of custom computer chips
powering it, and this could be one of them.â Neuman adds that
robomorphic computing might allow robots to relieve humans of risk
in a range of settings, such as caring for covid-19 patients or
manipulating heavy objects.
âThis work is exciting because it shows how specialized
circuit designs can be used to accelerate a core component of robot
control,â says Robin Deits, a robotics engineer at Boston
Dynamics who was not involved in the research. âSoftware
performance is crucial for robotics because the real world never
waits around for the robot to finish thinking.â He adds that
Neumanâs advance could enable robots to think faster,
âunlocking exciting behaviors that previously would be too
computationally difficult.â
Neuman next plans to automate the entire system of robomorphic
computing. Users will simply drag and drop their robotâs
parameters, and âout the other end comes the hardware
description. I think thatâs the thing thatâll push it over the
edge and make it really useful.â
This research was funded by the National Science Foundation, the
Computing Research Agency, the CIFellows Project, and the Defense
Advanced Research Projects Agency.
Originally published by
Daniel Ackerman | MIT News
Office | January 21, 2021
MIT