· Policy & regulation· 日本語で読む
NIST Proposes Benchmark to Assess Humanoid Robots’ Basic Capabilities
NIST has proposed a benchmark to assess the minimum physical capabilities of humanoid robots.
What happened
On April 20, 2026, NIST announced a proposal for a Baseline Performance Benchmark to assess humanoid robots’ basic capabilities. Based on tests NIST designed for the 2013-2014 DARPA Robotics Challenge (DRC), the benchmark consists of low-footprint locomotion and manipulation tasks derived largely from existing standardized NIST test methods. Designs and 3D models of the test apparatus will be made public. Manufacturers can choose to receive the apparatus and conduct their own tests or have their robots tested at NIST or participating facilities. To protect intellectual property, test results will be collected under preapproved data-sharing agreements between NIST and manufacturers, and aggregate results will be published to show the current state of humanoid capabilities.
Why it matters
For humanoid robot developers and researchers, the benchmark offers a way to test, under conditions aligned with existing standardized tests, the basic locomotion and object-manipulation capabilities expected of commercial robots intended for factories, homes and health care. NIST says the apparatus designs and 3D models will be public and can be used not only for physical testing but also for training and control development, making it easier for companies and institutions to track progress against the same set of tasks.
Sources
Translated from the Japanese original: /news/nist-hyuumanoidorobotto-no-beesurain-seinou-benchimaaku-wo-k-33f709. The Japanese version is authoritative if they differ.