Multi-Momentum Observer Contact Estimation for Bipedal Robots

TL;DR

This work addresses robust contact-mode estimation for bipedal robots without ground-contact sensors by introducing a multi-momentum-observer framework. It constructs constrained, reduced-order dynamics for each potential contact mode, runs a momentum observer per mode, uses relative-foot velocity as a liftoff cue, and fuses mode likelihoods with a Markov model to infer the active contact state. The approach achieves high accuracy in simulation ($\approx 98.4\%$) and substantial performance in hardware ($\approx 77.1\%$), demonstrating reliable proprioceptive contact estimation in large-scale bipedal systems. This enables more robust control for walkers and exoskeletons without hardware modifications, with potential extensions to richer contact modes and flight-like behaviors.

Abstract

As bipedal robots become more and more popular in commercial and industrial settings, the ability to control them with a high degree of reliability is critical. To that end, this paper considers how to accurately estimate which feet are currently in contact with the ground so as to avoid improper control actions that could jeopardize the stability of the robot. Additionally, modern algorithms for estimating the position and orientation of a robot's base frame rely heavily on such contact mode estimates. Dedicated contact sensors on the feet can be used to estimate this contact mode, but these sensors are prone to noise, time delays, damage/yielding from repeated impacts with the ground, and are not available on every robot. To overcome these limitations, we propose a momentum observer based method for contact mode estimation that does not rely on such contact sensors. Often, momentum observers assume that the robot's base frame can be treated as an inertial frame. However, since many humanoids' legs represent a significant portion of the overall mass, the proposed method instead utilizes multiple simultaneous dynamic models. Each of these models assumes a different contact condition. A given contact assumption is then used to constrain the full dynamics in order to avoid assuming that either the body is an inertial frame or that a fully accurate estimate of body velocity is known. The (dis)agreement between each model's estimates and measurements is used to determine which contact mode is most likely using a Markov-style fusion method. The proposed method produces contact detection accuracy of up to 98.44% with a low noise simulation and 77.12% when utilizing data collect on the Sarcos Guardian XO robot (a hybrid humanoid/exoskeleton).

Figures (7)

• Figure 1: Left: The hybrid humanoid robot/exoskeleton on which we demonstrate these algorithms. Right: The estimated external torques on the system during a gait cycle using each separate momentum observer.
• Figure 2: The five link walker model we use in simulation to evaluate the algorithm. The floating base location is defined as a position ($x_b,y_b$) and angle ($\theta_b$) relative to the vertical. The link angles are defined relative to straight hip ($\theta_1,\theta_3$) and straight knee ($\theta_2,\theta_4$) positions.
• Figure 3: A confusion matrix for the mode estimation from the simulated five link walker over a five second trajectory with a signal to noise ratio of 70dB. The values on the diagonal in blue represent correct estimations and the values off the diagonal in red represent incorrect estimations. The bottom and side contain the percentages of the time that a specific estimate is correct and that a specific mode is correctly estimated, respectively.
• Figure 4: A comparison between the current contact mode estimate and the ground truth of the simulated five link biped system on a trial with a signal to noise ratio of 70dB. The estimate is shown as a blue dashed line and the ground truth is shown as a solid yellow line. An inset highlights the short lag in detecting touchdown and liftoff at a stance transition event.
• Figure 5: Estimation accuracy as compared to the signal to noise ratio of the system. With lower noise levels, we see very accurate mode estimation. However, as the momentum observer is sensitive to velocity noise, the performance degrades notably with higher noise levels.