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

Developing robust vision-guided controllers for quadrupedal robots in complexenvironments, with various obstacles, dynamical surroundings and uneventerrains, is very challenging. While Reinforcement Learning (RL) provides apromising paradigm for agile locomotion skills with vision inputs insimulation, it is still very challenging to deploy the RL policy in the realworld. Our key insight is that aside from the discrepancy in the domain gap, invisual appearance between the simulation and the real world, the latency fromthe control pipeline is also a major cause of difficulty. In this paper, wepropose Multi-Modal Delay Randomization (MMDR) to address this issue whentraining RL agents. Specifically, we simulate the latency of real hardware byusing past observations, sampled with randomized periods, for bothproprioception and vision. We train the RL policy for end-to-end control in aphysical simulator without any predefined controller or reference motion, anddirectly deploy it on the real A1 quadruped robot running in the wild. Weevaluate our method in different outdoor environments with complex terrains andobstacles. We demonstrate the robot can smoothly maneuver at a high speed,avoid the obstacles, and show significant improvement over the baselines. Ourproject page with videos is at https://mehooz.github.io/mmdr-wild/.

Vision-Guided Quadrupedal Locomotion in the Wild with Multi-Modal Delay Randomization