English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This work presents a hierarchical framework for bipedal locomotion thatcombines a Reinforcement Learning (RL)-based high-level (HL) planner policy forthe online generation of task space commands with a model-based low-level (LL)controller to track the desired task space trajectories. Different fromtraditional end-to-end learning approaches, our HL policy takes insights fromthe angular momentum-based linear inverted pendulum (ALIP) to carefully designthe observation and action spaces of the Markov Decision Process (MDP). Thissimple yet effective design creates an insightful mapping between alow-dimensional state that effectively captures the complex dynamics of bipedallocomotion and a set of task space outputs that shape the walking gait of therobot. The HL policy is agnostic to the task space LL controller, whichincreases the flexibility of the design and generalization of the framework toother bipedal robots. This hierarchical design results in a learning-basedframework with improved performance, data efficiency, and robustness comparedwith the ALIP model-based approach and state-of-the-art learning-basedframeworks for bipedal locomotion. The proposed hierarchical controller istested in three different robots, Rabbit, a five-link underactuated planarbiped; Walker2D, a seven-link fully-actuated planar biped; and Digit, a 3Dhumanoid robot with 20 actuated joints. The trained policy naturally learnshuman-like locomotion behaviors and is able to effectively track a wide rangeof walking speeds while preserving the robustness and stability of the walkinggait even under adversarial conditions.

Template Model Inspired Task Space Learning for Robust Bipedal Locomotion