NIST proposes a baseline performance benchmark for humanoid robots

NIST proposes a baseline performance benchmark for humanoid robots

Rendering of a proposed equipment for standardized testing of humanoid robotic capabilities. Supply: NIST and ChatGPT

As the worldwide race to a commercially profitable humanoid robotic continues, the necessity to have the ability to consider them is rising. The Nationwide Institute for Requirements and Know-how, or NIST, final month proposed a “complete methodology to judge minimal anticipated bodily capabilities for humanoid robots.”

“NIST is proposing the primary standardized efficiency benchmark for humanoid robots for the reason that 2015 DARPA Robotics Problem,” famous Aaron Prather, director of the Robotics & Autonomous Methods Program at ASTM Worldwide, on LinkedIn. “In a decade that’s seen [Tesla‘s] Optimus, Determine, Agility, Apptronik, Unitree, and a dozen different humanoid platforms entice billions in funding, there’s nonetheless no agreed-upon solution to measure what any of them can truly do. Advertising and marketing movies have crammed the hole.”

Bethesda, Md.-based NIST’s Intelligent Systems Division mentioned it constructed the proposal on its earlier collaboration with the Protection Superior Analysis Initiatives Company (DARPA) to judge humanoid capabilities throughout {industry} and academia and to supply path for future growth. The institute is a part of the U.S. Division of Commerce.

Earlier this month, Fraunhofer IPA shared its personal benchmark for humanoid security and growth with six standards.

NIST to check humanoid locomotion and manipulation

NIST described its proposed baseline efficiency benchmark as “a low-footprint set of locomotion and manipulation duties.” They use beforehand outlined and standardized take a look at strategies and efficiency metrics.

The duties are meant to symbolize the minimal capabilities for commercially obtainable humanoid robots in industrial, family, healthcare, or different purposes. NIST mentioned the outcomes will set up functionality measurements for the present industry-leading robots, in addition to a standard set of duties for researchers and producers.

The group added that the duties will present a baseline for efficiency whereas additionally testing:

  • Area-agnostic, fundamental humanoid robotic mobility and manipulation/dexterity capabilities
  • Coordinated capabilities combining locomotion and manipulation duties
  • Entire-body consciousness and management by confined-space manipulation duties
  • Minimal reasoning, job and scene understanding, and resolution making
Schematic of an apparatus for testing humanoid capabilities from NIST.

NIST has developed an equipment for testing humanoid capabilities. (Click on right here to enlarge.) Supply: NIST

Benchmarking course of the results of collaboration

NIST mentioned it has designed the baseline benchmark equipment in collaboration with {industry} and the analysis neighborhood. It’s looking for contributors because it develops a consensus on what duties ought to be included.

The institute plans to construct a restricted variety of these testing apparatuses totally free distribution to U.S. humanoid robotic producers and established regional testing services. NIST additionally plans to publish the designs and 3D fashions of the equipment to be used as a bodily and/or digital testbed for robotic coaching and management growth.

Robotic producers that select to check their robots may obtain an equipment (pending availability) to run their very own assessments, or they’ll take a look at their robots at NIST or a taking part facility. The humanoid makers and NIST will accumulate the take a look at outcomes underneath pre-approved data-sharing agreements to guard mental property and any attribution if wanted.

NIST plans to mixture outcomes to showcase the state-of-the-art of humanoid robotic capabilities.

Task list for NIST humanoid benchmark testing.

Activity listing for humanoid benchmark testing. (Click on right here to enlarge.) Supply: NIST

Issues for potential contributors

  1. Do these duties sufficiently train minimal humanoid capabilities?
  2. Are there duties that will higher train whole-body-control and/or loco-manipulation?
  3. What extra constraints would you want for testing your robots in such a collection of duties?
  4. Are you keen to assist design the benchmark and/or have your robots examined?
  5. Are you interested by being a taking part take a look at facility?

To be taught extra or to take part, NIST has a contact form. events can even contact Dr. Benjamin Beiter at benjamin.beiter@nist.gov and Dr. Kamel Saidi at kamel.saidi@nist.gov.

The put up NIST proposes a baseline efficiency benchmark for humanoid robots appeared first on The Robotic Report.