Cassie, bipedal robot, Dynamic Robotics Laboratory, Corvallis, Oregon, USA

[Administrator]

Article "Agility Robotics' Cassie Is Now Astonishingly Good at Stairs"
Using only proprioceptive sensors, this bipedal robot is probably better at stairs than you are

by Evan Ackerman
May 20, 2021

 

[Administrator]

Fully autonomy on the Wave Field 2021

Jun 4, 2021

During the dark of night, using LiDAR for eyes, Cassie Blue is operating fully autonomously on the University of Michigan Wave Field. The terrain is challenging and was not pre-mapped.


The LiDAR and IMU data are fused in real-time to form an elevation map. A CLF- RRT* planner is running at 5Hz with reactive re-planning performed at 300 Hz on the basis of a Control Lyapunov Function (CLF). The reactive planner provides velocity commands to a One-step Ahead Gait Controller based on Angular Momentum https://youtu.be/V36DCsc6iio . The paper https://arxiv.org/abs/2105.08170 provides full details on the controller. The planner is currently unpublished.

This work was presented at the 5th ICRA Workshop on Legged Robotics on 4 June 2021.
docs.google.com/presentation/d/1onC0VqRHnzqb3jgufmPRgeIwla74jQLFmeLGrjASdRw/edit#slide=id.g4cee01c7e5_0_24

 

[Administrator]

Jul 27, 2021

To encourage us to push the limits of reliability, energy efficiency and speed we attempted to run a 5k (5 kilometers, 3.1 miles) with our bipedal robot Cassie.

Our final goal was to run a 5k through the Oregon State University campus. That can be seen here https://youtu.be/dY57qnD_O7U

This turf 5k was a test before we attempted the campus run. We finished with a time of 43:59.53 which means we had an average speed of 1.89 m/s. This is the fastest bipedal robot 5k at the time of posting.

Machine learning breakthrough: Robot runs a 5k

Jul 27, 2021

The OSU Dynamic Robotics Laboratory's research team, led by Agility Robotics’ Jonathan Hurst, combined expertise from biomechanics and robot controls with new machine learning tools to accomplish something new: train a bipedal robot to run a full 5K on a single battery charge! This industry-first invention will unleash new levels of robot performance. Today, some of these same researchers are Agility Robotics employees busily applying their know-how to Digit, so just wait to see what we have in store.

"Cassie the bipedal robot runs a 5K"

by Brian Heater
July 28, 2021

 

[Administrator]

Terrain-Aware Foot Placement for Bipedal Locomotion Combining MPC, Virtual Constraints, and the ALIP

Nov 4, 2021

Title: Terrain-Aware Foot Placement for Bipedal Locomotion Combining Model Predictive Control, Virtual Constraints, and the ALIP

Authors: Grant Gibson, Oluwami Dosunmu-Ogunbi, Yukai Gong, Jessy Grizzle

Preprint link: https://arxiv.org/abs/2109.14862

Abstract:
This paper draws upon three themes in the bipedal control literature to achieve highly agile, terrain-aware locomotion. By terrain aware, we mean the robot can use information on terrain slope and friction cone as supplied by state-of-the-art mapping and trajectory planning algorithms. The process starts with abstracting from the full dynamics of a Cassie 3D bipedal robot, an exact low-dimensional representation of its centroidal dynamics, parameterized by angular momentum. Under a piecewise planar terrain assumption, and the elimination of terms for the angular momentum about the robot's center of mass, the centroidal dynamics become linear and has dimension four. Four-step-horizon model predictive control (MPC) of the centroidal dynamics provides step-to-step foot placement commands. Importantly, we also include the intra-step dynamics at 10 ms intervals so that realistic terrain-aware constraints on robot's evolution can be imposed in the MPC formulation. The output of the MPC is directly implemented on Cassie through the method of virtual constraints. In experiments, we validate the performance of our control strategy for the robot on inclined and stationary terrain, both indoors on a treadmill and outdoors on a hill.

We thank Margaret Eva Mungai, Jennifer Humanchuk, and Jianyang Tang for assistance in experiments.

 

[Administrator]

Cassie autonomously navigates around obstacles

Nov 17, 2021

Cassie Blue navigates around furniture treated as obstacles in the atrium of the Ford Robotics Building at the University of Michigan.

 

[Administrator]

Cassie autonomously navigates in four long corridors (200 meters)

Nov 29, 2021

Cassie Blue autonomously navigates on the second floor of the Ford Robotics Building at the University of Michigan. The total traverse distance is 200 m (656.168 feet).

The LiDAR and IMU data are fused in real-time to form an elevation map. The system consists of a low-frequency planning thread (5 Hz) to find an asymptotically optimal path and a high-frequency reactive thread (300 Hz) to accommodate robot deviation. The planning thread includes: a multi-layer local map to compute traversability for the robot on the terrain; an anytime omnidirectional Control Lyapunov Function (CLF) for use with a Rapidly Exploring Random Tree Star (RRT*) that generates a vector field for specifying motion between nodes; a sub-goal finder when the final goal is outside of the current map; and a finite-state machine to handle high-level mission decisions. The paper [ https://arxiv.org/abs/2108.06699 ] provides full details on the reactive planning system.

 

[Administrator]

Terrain - Adaptive, ALIP-based bipedal locomotion controller via MPC and virtual constraints - extended

Jul 30, 2022

Extended Video for IROS 2022 Accepted Paper (https://arxiv.org/abs/2109.14862)

Authors: Grant Gibson, Oluwami Dosunmu-Ogunbi, Yukai Gong, and Jessy Grizzle

Title: Terrain-Adaptive, ALIP-Based Bipedal Locomotion Controller via Model Predictive Control and Virtual Constraints

Abstract:
This paper presents a gait controller for bipedal robots to achieve highly agile walking over various terrains given local slope and friction cone information. Without these considerations, untimely impacts can cause a robot to trip and inadequate tangential reaction forces at the stance foot can cause slippages. We address these challenges by combining, in a novel manner, a model based on an Angular Momentum Linear Inverted Pendulum (ALIP) and a Model Predictive Control (MPC) foot placement planner that is executed by the method of virtual constraints. The process starts with abstracting from the full dynamics of a Cassie 3D bipedal robot, an exact low-dimensional representation of its center of mass dynamics, parameterized by angular momentum. Under a piecewise planar terrain assumption and the elimination of terms for the angular momentum about the robot's center of mass, the centroidal dynamics about the contact point become linear and have dimension four. Importantly, we include the intra-step dynamics at uniformly-spaced intervals in the MPC formulation so that realistic workspace constraints on the robot's evolution can be imposed from step-to-step. The output of the low-dimensional MPC controller is directly implemented on a high-dimensional Cassie robot through the method of virtual constraints. In experiments, we validate the performance of our control strategy for the robot on a variety of surfaces with varied inclinations and textures.

 

[Administrator]

Cassie sets world record for 100M run

Sep 27, 2022

Cassie, the robot, clocked the historic time of 24.73 seconds at the Whyte Track and Field Center, starting from a standing position and returning to that position after the sprint, with no falls. This incredible achievement was accomplished through robot learning and almost a year of simulation, condensed down to a matter of weeks.

Cassie was invented at the Oregon State University College of Engineering and produced by Agility Robotics. Through this work Cassie has established a Guinness World Record for the fastest 100 meters by a bipedal robot.

 

[Administrator]

Cassie on a Moving Walkway

Nov 21, 2023

 

[Administrator]

Cassie walks on sand, gravel, and rocks in the Robot Playground

Apr 1, 2024

Using the controller from the paper below, Cassie is able to walk on sand, gravel, and rocks inside the Robot Playground at the University of Michigan

Paper: https://arxiv.org/abs/2403.02486

 

[Administrator]

Learning vision-based bipedal locomotion for challenging terrain

Apr 6, 2024

Supplementary video for 2024 IEEE International Conference on Robotics and Automation
arXiv: https://arxiv.org/abs/2309.14594

Abstract - Reinforcement learning (RL) for bipedal locomotion has recently demonstrated robust gaits over moderate terrains using only proprioceptive sensing. However, such blind controllers will fail in environments where robots must anticipate and adapt to local terrain, which requires visual perception. In this paper, we propose a fully-learned system that allows bipedal robots to react to local terrain while maintaining commanded travel speed and direction. Our approach first trains a controller in simulation using a heightmap expressed in the robot's local frame. Next, data is collected in simulation to train a heightmap predictor, whose input is the history of depth images and robot states. We demonstrate that with appropriate domain randomization, this approach allows for successful sim-to-real transfer with no explicit pose estimation and no fine-tuning using real-world data. To the best of our knowledge, this is the first example of sim-to-real learning for vision-based bipedal locomotion over challenging terrains.