KORTHOS / ECOSYSTEM / PRODUCT

ATRIAS

About

About ATRIAS

Series-elastic leg assemblies and a torso-centred mass distribution gave the machine dynamics modelled on a spring-loaded inverted pendulum. Researchers used onboard power and computing to run untethered controllers, measure energy-efficient gaits and study recovery when terrain or external disturbances disrupted the planned motion.

Multiple laboratory builds supported walking, running and field experiments before the programme gave way to Cassie. The work informed technology transferred from Oregon State's Dynamic Robotics Laboratory into the company that became Agility Robotics. ATRIAS is retained as research history and lineage evidence; it is not offered in Agility's current catalogue and should not be presented as a commercial deployment platform.

Capabilities

Source-backed capabilities

  • Capability areas: Bipedal locomotion
  • Workflows: Dynamic walking research, legged locomotion validation, control research and bipedal robot development.
  • Operating environments: University labs, locomotion research environments and robotics history/context pages.
  • Target markets: Researchers and readers tracking the technical lineage behind Agility Robotics platforms.
  • Deployment model: Research-platform model rather than a current commercial robot.
  • Bipedal walking

    Research demonstration - 2017-09-18 - The paper applies Bayesian optimization with a domain-informed feature transform to bipedal locomotion controller tuning.

    Source

Research

  • Deep-RL policy-transfer paper reports hardware results on ATRIAS

    2018-09-28 - Research Publication - The paper studies whether deep reinforcement-learning policies can transfer from simulation to hardware when embedded inside a structured walking controller. Experiments on ATRIAS examine action spaces, cost functions, and the rate of transfer from high-fidelity simulation to the physical biped. The result is a historically important ATRIAS entry for combining learned policies with interpretable controller structure.

    Source
  • Bayesian-optimization study tunes walking controllers on ATRIAS hardware

    2017-09-18 - Research Publication - The paper applies Bayesian optimization with a domain-informed feature transform to bipedal locomotion controller tuning. The method is evaluated on the ATRIAS biped in simulation and on hardware, including two walking controllers on the real robot. It gives ATRIAS a concrete learning-on-hardware research record for sample-efficient bipedal gait optimization.

    Source

Context

Market context

ATRIAS is the bipedal research platform behind key dynamic walking work that preceded Cassie and Digit.

  • Target environment: University labs, locomotion research environments and robotics history/context pages.
  • Workflow context: Dynamic walking research, legged locomotion validation, control research and bipedal robot development.
  • Customer context: Researchers and readers tracking the technical lineage behind Agility Robotics platforms.
  • Deployment model: Research-platform model rather than a current commercial robot.
  • Commercial maturity: Historical research platform with strong technical relevance.
  • Adoption constraints: Adoption is historical; relevance depends on locomotion lineage and research context.
  • Market position: Foundational bipedal locomotion platform connected to later Agility Robotics products.
  • Adjacent products: Cassie, Digit, Oregon State Dynamic Robotics Lab work

Tags

  • Capability: Bipedal locomotion
  • Commercial Stage: Research platform

Comparable Systems