University of Illinois Develops CARL for Enhanced Robot Trajectory Tracking by 15%
- Published
- Source
- InterestingEngineering.com
Researchers at the University of Illinois Urbana-Champaign have introduced a novel learning method that enhances robot trajectory tracking performance by up to 15%. This method, known as Contraction-Aware Reinforcement Learning (CARL), integrates reinforcement learning with control theory to ensure stability while improving task execution across various robotic platforms, including wheeled, flying, and six-legged robots.
The significance of CARL lies in its ability to make learning-based robotic systems more practical by minimizing the risks of instability and potential damage during operation. By providing mathematical guarantees on tracking performance, CARL allows robots to follow desired paths more reliably, addressing a common challenge in reinforcement learning where robots may exhibit unpredictable behavior.
Looking ahead, the research team has successfully tested CARL on multiple robotic platforms, including a wheeled TurboBot 3 Burger, a flapping aerial drone, and a six-legged Jethexa robot. No further timeline was disclosed at the time of publication, but the implications of this research could lead to broader applications of stable, learning-based control in robotics.
Original report
Robots get 15% better trajectory tracking for flying and walking with new framework
InterestingEngineering.com · Jijo MalayilThis briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
Related Suppliers
Sierra NV Corp
United States
US company developing autonomous off-road vehicles and defense ground vehicle systems.
Deci AI (Acquired by NVIDIA)
Israel
Israeli deep learning optimization startup acquired by NVIDIA in May 2024 for ~$300M; known for AutoNAC neural architecture search technology.
GIPSA-lab (UMR 5216)
France
GIPSA-lab is a Grenoble signal, control and speech lab (UMR 5216) of 320 members that builds custom flying robots and the NINA humanoid robot.
Northeastern University — Institute for Experiential Robotics
United States
Boston-based Northeastern University institute at 815 Columbus Avenue; researches legged, soft and medical robotics through labs led by Lipton, Padir and Song.
Reward AI
United States
Bay Area robot-learning startup founded 2025; its OM-1 policy learns manipulation only from human demos captured with a 7-DOF wearable hand.
BabaCAD
Bosnia and Herzegovina
Bosnia and Herzegovina CAD developer behind BabaCAD Robotics offline programming and a free browser robot simulator; its CAD cites 250,000+ users.
KAIST Hubo Lab (Humanoid Robot Research Center)
Korea, Republic of (South)
KAIST humanoid research centre that built HUBO and won the 2015 DARPA Robotics Challenge with DRC-HUBO+; it now develops a 3.25 m/s next-gen humanoid.
Raion Robotics
Korea, Republic of (South)
Daejeon, South Korea KAIST spin-off founded 2023; builds the Raibo2 quadruped, which ran a full marathon on one charge in 2024.
Guangdong Ability Robotics Co., Ltd. 广东爱比力智能科技有限公司
China
Foshan maker of rehabilitation robots: SuiHeng gait and balance robot, YiBu body-weight-support trainer and the SuiHeng Pro hip exoskeleton
Holiday Robotics
Korea, Republic of (South)
South Korean startup building FRIDAY, a wheeled 64-DoF dexterous industrial humanoid, after a $105M Series A round in 2026.
GMO AI & Robotics Corporation
Japan
Tokyo-area GMO AI & Robotics, founded June 2024 as a GMO Internet Group unit, is Japan's authorized Unitree Robotics distributor for humanoid robots.
Telefactor Robotics
United States
US specialist in teleoperated robot vision, dexterous grippers, and control interfaces for bomb disposal and hazardous environment robots.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.
