Cassie

Source-backed capabilities

• Workflows: Bipedal locomotion research, controller development, robot learning, mobility validation and platform experimentation.
• Operating environments: Academic labs, industrial R&D sites, locomotion research programs and robot-learning environments.
• Target markets: Researchers, universities and industrial R&D teams studying legged robot mobility.
• Deployment model: Commercial research-platform model used before Agility's warehouse-focused Digit product.
• Dynamic locomotion

  Research demonstration - 2024-01-30 - The IJRR study presents a reinforcement-learning controller for standing, walking, running, jumping, and other dynamic bipedal skills.

  Source
• Multi-terrain locomotion

  Research demonstration - 2023-09-14 - The work gives Cassie a benchmark for rough-terrain footstep planning with small feet and limited ankle authority.

  Source
• Bipedal walking

  Research demonstration - 2023-09-14 - The paper introduces a mixed-integer model-predictive footstep controller for bipedal walking over constrained footholds.

  Source
• Dynamic locomotion

  Research demonstration - 2023-02-19 - The RSS paper trains a reinforcement-learning controller for robust bipedal jumping across different distances and directions.

  Source
• Bipedal walking

  Research demonstration - 2021-09-30 - The work gives Cassie a benchmark for agile bipedal walking over varied terrain conditions.

  Source
• Obstacle-course traversal

  Research demonstration - 2019-07-26 - Experiments include flat terrain, slopes, stairs, rough outdoor terrain, and autonomous obstacle-course navigation using visual SLAM and planning.

  Source
• Bipedal walking

  Research demonstration - 2018-09-19 - The work gives Cassie an early benchmark as a shared academic platform for bipedal locomotion and control research.

  Source
• SLAM mapping

  Research demonstration - 2019-07-26 - Experiments include flat terrain, slopes, stairs, rough outdoor terrain, and autonomous obstacle-course navigation using visual SLAM and planning.

  Source
• Multi-vehicle coordination

  Research demonstration - 2024-06-25 - It adds a distinctive Cassie research result around multi-robot coordination and legged payload transport over terrain.

  Source
• Warehouse automation

  Demonstrated - 2022-09-27 - US robotics company focused on humanoid systems for logistics and warehouse automation.

  Source
• Load carrying

  Research demonstration - 2024-06-25 - The paper trains a decentralized controller for payload transport using multiple rigidly attached biped robots.

  Source

• SAC-Based Gait Generation and Robust Balance Control for Cassie Robots in Unknown Terrain

  2026-02-15 - Research Publication - The development of robust locomotion strategies for bipedal robots remains a significant challenge in robotics, particularly when navigating unstructured and unknown environments. The Cassie robot, a dynamic bipedal platform with high degrees of freedom and underactuated passive dynamics, presents specific control difficulties that classical model-based approaches often fail to address adequately due to modeling mismatches and computational latency. This paper proposes a novel framework utilizing Soft Actor-Critic (SAC), an off-policy deep reinforcement learning algorithm, to generate stable gait patterns and ensure robust balance control. Unlike standard on-policy methods, SAC optimizes a maximum entropy objective, which encourages substantial exploration and provides greater robustness to external disturbances. We introduce a comprehensive reward function design and a domain randomizat

  Source
• Multi-biped payload-transport paper demonstrates coordinated Cassie teams

  2024-06-25 - Research Publication - Payload transport over flat terrain via multi-wheel robot carriers is well-understood, highly effective, and configurable.

  Source
• IJRR study demonstrates broad dynamic locomotion skills on Cassie

  2024-01-30 - Research Publication - This paper covers IJRR study demonstrates broad dynamic locomotion skills on Cassie for Cassie in Reinforcement learning.

  Source
• Cassie constrained-foothold paper demonstrates MPC footstep control

  2023-09-14 - Research Publication - Bipedal robots promise the ability to traverse rough terrain quickly and efficiently, and indeed, humanoid robots can now use strong ankles and careful foot placement to traverse discontinuous terrain.

  Source
• RSS jumping-control paper transfers versatile dynamic jumps to Cassie hardware

  2023-02-19 - Research Publication - This paper covers RSS jumping-control paper transfers versatile dynamic jumps to Cassie hardware for Cassie in Robot learning.

  Source
• Leveraging Sequentiality in Reinforcement Learning from a Single Demonstration

  2022-11-09 - Research Publication - Deep Reinforcement Learning has been successfully applied to learn robotic control. However, the corresponding algorithms struggle when applied to problems where the agent is only rewarded after achieving a complex task.

  Source
• Value Summation: A Novel Scoring Function for MPC-based Model-based Reinforcement Learning

  2022-09-16 - Research Publication - This paper proposes a novel scoring function for the planning module of MPC-based reinforcement learning methods to address the inherent bias of using the reward function to score trajectories. The proposed method enhances the learning efficiency of existing MPC-based MBRL methods using the discounted sum of values.

  Source
• Learning Dynamic Bipedal Walking Across Stepping Stones

  2022-05-03 - Research Publication - In this work, we propose a learning approach for 3D dynamic bipedal walking when footsteps are constrained to stepping stones. While recent work has shown progress on this problem, real-world demonstrations have been limited to relatively simple open-loop, perception-free scenarios.

  Source
• Custom Sine Waves Are Enough for Imitation Learning of Bipedal Gaits with Different Styles

  2022-04-08 - Research Publication - Not until recently, robust bipedal locomotion has been achieved through reinforcement learning. However, existing implementations rely heavily on insights and efforts from human experts, which is costly for the iterative design of robot systems.

  Source
• Cassie ALIP-MPC paper demonstrates terrain-adaptive bipedal locomotion

  2021-09-30 - Research Publication - This paper presents a gait controller for bipedal robots to achieve highly agile walking over various terrains given local slope and friction cone information.

  Source
• Control Lyapunov Functions for Compliant Hybrid Zero Dynamic Walking

  2021-07-09 - Research Publication - The ability to realize nonlinear controllers with formal guarantees on dynamic robotic systems has the potential to enable more complex robotic behaviors -- yet, realizing these controllers is often practically challenging. To address this challenge, this paper presents the end-to-end realization of dynamic bipedal locomotion on an underactuated bipedal robot via hybrid zero dynamics and control Lyapunov functions.

  Source
• Parameterized locomotion RL paper transfers robust walking policies to Cassie

  2021-03-26 - Research Publication - Developing robust walking controllers for bipedal robots is a challenging endeavor. Traditional model-based locomotion controllers require simplifying assumptions and careful modelling; any small errors can result in unstable control.

  Source
• Cassie sim-to-real paper learns all common bipedal gaits

  2020-11-02 - Research Publication - We study the problem of realizing the full spectrum of bipedal locomotion on a real robot with sim-to-real reinforcement learning (RL).

  Source
• Velocity Regulation of 3D Bipedal Walking Robots with Uncertain Dynamics\n Through Adaptive Neural Network Controller

  2020-08-01 - Research Publication - This paper presents a neural-network based adaptive feedback control\nstructure to regulate the velocity of 3D bipedal robots under dynamics\nuncertainties. Existing Hybrid Zero Dynamics (HZD)-based controllers regulate\nvelocity through the implementation of heuristic regulators that do not\nconsider model and environmental uncertainties, which may significantly affect\nthe tracking performance of the controllers. In this paper, we address the\nuncertainties in the robot dynamics from the perspective of the reduced\ndimensional representation of virtual constraints and propose the integration\nof an adaptive neural network-based controller to regulate the robot velocity\nin the presence of model parameter uncertainties. The proposed approach yields\nimproved tracking performance under dynamics uncertainties. The shallow\nadaptive neural network used in this paper does not require traini

  Source
• Cassie Hovershoes paper demonstrates multimodal bipedal riding control

  2019-07-26 - Research Publication - Motivated towards achieving multi-modal locomotion, in this paper, we develop a framework for a bipedal robot to dynamically ride a pair of Hovershoes over various terrain.

  Source
• Cassie feedback-control paper demonstrates walking, standing, and Segway riding

  2018-09-19 - Research Publication - The Cassie bipedal robot designed by Agility Robotics is providing academics a common platform for sharing and comparing algorithms for locomotion, perception, and navigation.

  Source

• Agility Robotics sets world record with Cassie's learned policy

  2022-09-27 - Demonstration - This achievement highlights advancements in robotic learning and performance capabilities. Cassie's record-setting moment showcases the potential for robotics in competitive environments. US robotics company focused on humanoid systems for logistics and warehouse automation.

  Source
• Cassie the bipedal robot completes 5K run

  2021-07-25 - Demonstration - Cassie, a bipedal robot developed at Oregon State University and produced by Agility Robotics, completed a 5-kilometer run in just over 53 minutes. The run took place on Oregon State's campus, with Cassie operating untethered and on a single battery charge. During the run, Cassie experienced two falls: one due to an overheated computer and another from executing a turn at too high a speed.

  Source
• Agility Robotics delivers Cassie 001 to University of Michigan

  2017-09-05 - Market Signal - Agility Robotics delivered the first Cassie bipedal robot, serial number 001, to the University of Michigan on September 5, 2017. The university named the robot 'Cassie Blue' and began testing it on various terrains, including sidewalks, grass, snow, sand, and burning brush. The robot featured two extra motors in each leg for enhanced mobility.

  Source
• Agility Robotics completes first Cassie robot deliveries

  2017-08-31 - Market Signal - Agility Robotics delivered its first batch of Cassie robots to the University of Michigan, Caltech, and Harvard at the end of August and beginning of September 2017. Each team began developing its own control laws for the robots upon receipt. The deliveries marked the initial deployment of Cassie robots to academic institutions.

  Source

Market context

Cassie is Agility Robotics' first commercial bipedal robot platform, derived from ATRIAS research and used for R&D before Digit.

• Target environment: Academic labs, industrial R&D sites, locomotion research programs and robot-learning environments.
• Workflow context: Bipedal locomotion research, controller development, robot learning, mobility validation and platform experimentation.
• Customer context: Researchers, universities and industrial R&D teams studying legged robot mobility.
• Deployment model: Commercial research-platform model used before Agility's warehouse-focused Digit product.
• Commercial maturity: Established research and development platform.
• Adoption constraints: Adoption depends on research goals, support, simulation tools and platform availability.
• Market position: Bridge product between university bipedal locomotion research and deployable Agility Robotics systems.
• Adjacent products: ATRIAS, Digit, Unitree H1, Boston Dynamics Atlas

Comparable Systems

• TRON 1

  LimX Dynamics - Research Platform

• ATRIAS

  Agility Robotics - Research Platform

• NASA/JPL CADRE

  NASA - Space Robot