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

An agent's functionality is largely determined by its design, i.e., skeletalstructure and joint attributes (e.g., length, size, strength). However, findingthe optimal agent design for a given function is extremely challenging sincethe problem is inherently combinatorial and the design space is prohibitivelylarge. Additionally, it can be costly to evaluate each candidate design whichrequires solving for its optimal controller. To tackle these problems, our keyidea is to incorporate the design procedure of an agent into itsdecision-making process. Specifically, we learn a conditional policy that, inan episode, first applies a sequence of transform actions to modify an agent'sskeletal structure and joint attributes, and then applies control actions underthe new design. To handle a variable number of joints across designs, we use agraph-based policy where each graph node represents a joint and uses messagepassing with its neighbors to output joint-specific actions. Using policygradient methods, our approach enables joint optimization of agent design andcontrol as well as experience sharing across different designs, which improvessample efficiency substantially. Experiments show that our approach,Transform2Act, outperforms prior methods significantly in terms of convergencespeed and final performance. Notably, Transform2Act can automatically discoverplausible designs similar to giraffes, squids, and spiders. Code and videos areavailable at https://sites.google.com/view/transform2act.

Transform2Act: Learning a Transform-and-Control Policy for Efficient Agent Design