Meta tags:
description=Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion;
keywords=Legged Locomotion, Agile Locomotion,Reinforcement Learning, Visuomotor Control, Sim-to-Real Transfer, Collision Avoidance;

Headings (most frequently used words):

collision, with, outdoor, test, agile, but, safe, learning, free, high, speed, legged, locomotion, real, world, demos, video, story, abstract, narrow, corridor, indoor, hall, furnitures, grass, snow, stroller, avoidance, diverse, objects, agility, robustness, baseline, comparison, method, abs, framework, media, bibtex,

Text of the page (most frequently used words):
the (14), and (10), agile (9), policy (9), trained (5), learning (4), safe (4), collision (4), #legged (4), abs (4), but (3), free (3), high (3), speed (3), locomotion (3), network (3), values (3), recovery (3), obstacles (3), with (3), from (2), tairan (2), zhang (2), chong (2), xiao (2), wenli (2), guanqi (2), liu (2), changliu (2), shi (2), guanya (2), rss (2), 2024 (2), architecture (2), value (2), activated (2), track (2), twist (2), commands (2), that (2), lower (2), threshold (2), amidst (2), framework (2), predict (2), ray (2), velocity (2), avoid (2), agility (2), test (2), outdoor (2), video (2), paper (2), page, template, borrowed, robot, parkour, nerfies, inproceedings, he2024agile, author, title, booktitle, robotics, science, systems, year, bibtex, media, deployment, dual, setup, switches, between, based, estimated, via, constrained, optimization, navigate, training, there, are, four, modules, within, distances, policies, exteroceptive, inputs, given, depth, images, prediction, desired, linear, yaw, angular, conditioned, safety, indicators, reach, achieve, maximum, method, lagrangian, methods, only, baseline, comparison, robustness, avoidance, diverse, objects, stroller, snow, grass, indoor, hall, furnitures, narrow, corridor, robots, navigating, cluttered, environments, must, jointly, for, efficient, task, execution, collisions, humans, existing, studies, either, develop, conservative, controllers, abstract, story, real, world, demos, code, summary, outstanding, student, award, finalist, top,

Text of the page (random words):
agile but safe learning collision free high speed legged locomotion agile but safe learning collision free high speed legged locomotion tairan he chong zhang wenli xiao guanqi he changliu liu guanya shi rss 2024 outstanding student paper award finalist top 3 paper video summary code real world demos video story abstract legged robots navigating cluttered environments must be jointly agile for efficient task execution and safe to avoid collisions with obstacles or humans existing studies either develop conservative controllers narrow corridor indoor hall with furnitures outdoor grass outdoor snow stroller collision avoidance with diverse objects agility test robustness test baseline comparison abs abs agile policy only lagrangian methods method abs framework training architecture there are four trained modules within the abs framework agile policy is trained to achieve the maximum agility amidst obstacles reach avoid value network is trained to predict the ra values conditioned on the agile policy as safety indicators recovery policy is trained to track desired twist commands 2d linear velocity and yaw angular velocity that lower the ra values ray prediction network is trained to predict ray distances as the policies exteroceptive inputs given depth images deployment architecture the dual policy setup switches between the agile policy and the recovery policy based on the estimated v̂ from the ra value network if v̂ threshold the agile policy is activated to navigate amidst obstacles if v̂ v threshold the recovery policy is activated to track twist commands that lower the ra values via constrained optimization media bibtex inproceedings he2024agile author he tairan and zhang chong and xiao wenli and he guanqi and liu changliu and shi guanya title agile but safe learning collision free high speed legged locomotion booktitle robotics science and systems rss year 2024 page template borrowed from nerfies and robot parkour learning