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

Recently, people have shown that large-scale pre-training from internet-scaledata is the key to building generalist models, as witnessed in NLP. To buildembodied generalist agents, we and many other researchers hypothesize that suchfoundation prior is also an indispensable component. However, it is unclearwhat is the proper concrete form to represent those embodied foundation priorsand how they should be used in the downstream task. In this paper, we proposean intuitive and effective set of embodied priors that consist of foundationpolicy, value, and success reward. The proposed priors are based on thegoal-conditioned MDP. To verify their effectiveness, we instantiate anactor-critic method assisted by the priors, called Foundation Actor-Critic(FAC). We name our framework as Foundation Reinforcement Learning (FRL), sinceit completely relies on embodied foundation priors to explore, learn andreinforce. The benefits of FRL are threefold. (1) Sample efficient. Withfoundation priors, FAC learns significantly faster than traditional RL. Ourevaluation on the Meta-World has proved that FAC can achieve 100% success ratesfor 7/8 tasks under less than 200k frames, which outperforms the baselinemethod with careful manual-designed rewards under 1M frames. (2) Robust tonoisy priors. Our method tolerates the unavoidable noise in embodied foundationmodels. We show that FAC works well even under heavy noise or quantizationerrors. (3) Minimal human intervention: FAC completely learns from thefoundation priors, without the need of human-specified dense reward, orproviding teleoperated demos. Thus, FAC can be easily scaled up. We believe ourFRL framework could enable the future robot to autonomously explore and learnwithout human intervention in the physical world. In summary, our proposed FRLis a novel and powerful learning paradigm, towards achieving embodiedgeneralist agents.

Foundation Reinforcement Learning: towards Embodied Generalist Agents with Foundation Prior Assistance