ICRA 2026 Competitions
ACCEPTED COMPETITIONS:
- AgiBot World Challenge 2026
- AI for Robotic Surgery
- Legged Robot Challenges
- LeHome Challenge 2026
- REAL-I: The 1st Real-world Embodied AI Learning Challenge
- SAI Soccer Challenge
- The 11th Robotic Grasping and Manipulation Competition
- The 27th RoboRacer Autonomous Racing Competition
- The BARN Challenge 2026
- What Bimanuals Can Do
AgiBot World Challenge
As embodied artificial intelligence evolves, the challenge shifts from mastering isolated skills to developing integrated systems capable of semantic reasoning, precise manipulation, and dynamic whole-body coordination. While previous benchmarks have laid the groundwork, they often separate high-level planning from low-level control. Building on the AgiBot World legacy, the 2026 ICRA competition introduces a comprehensive platform designed to evaluate the next generation of general-purpose humanoid robots in complex, unstructured environments.
This year’s competition emphasizes the convergence of “Brain” and “Body” through three distinct yet interconnected tracks: (1) The World Model Track, (2) The VLM + VLA Track, and (3) The Whole-Body Control (WBC) Track. By integrating these diverse challenges, our goal is to foster the development of robots that not only understand their surroundings through multi-modal data but can also physically interact with the world with agility and precision. We invite researchers to push the boundaries of embodied intelligence and shape the future of real-world robotic deployment.
AI for Robotic Surgery
Surgical robotics is entering an exciting new era where the integration of artificial intelligence (AI) promises to improve the performance of human surgeons, and to address the growing shortage of surgeons and other medical personnel, thereby improving healthcare for all.
The most prevalent surgical robot in operating rooms today is the da Vinci Surgical System (Intuitive Surgical, Sunnyvale, CA), which has an installed base of more than 10,000 systems. The competition will use the da Vinci Research Kit (dVRK), an open-source research platform that re-purposes the mechanical hardware from retired clinical da Vinci Surgical Systems. We will have two challenges, both using the peg transfer task, solving the task in a teleoperated and an autonomous manner.
Legged Robot Challenges
Autonomy in disaster response situations requires a good mobile platform, perception technologies, navigation technologies, etc. Legged Robot Challenges (LRC) is the extension of Quadruped Robot Challenges which are inherently dynamic.
Quadruped Robot Challenges (QRC) was inaugurated in ICRA 2023 to test teams for autonomous traversability on various kinds of terrains, with follow ups at ICRA 2024, IROS 2024, ICRA 2025, and IROS 2025. Recent surge of humanoid robot platforms has prepared us to extend QRC to LRC.
The LRC has great potential to lead the robotics community in technological advancement, inform researchers about the rigors required to reliably deploy into unstructured environments, and foster interactions with commercial manufacturers to create practical robotic systems.
LeHome Challenge 2026: Challenge on Garment Manipulation Skill Learning in Household Scenarios
Garment manipulation is a fundamental yet highly challenging problem in the robotic manipulation area, involving complex, deformable objects and contact-rich interactions. The Lehome Challenge aims to establish a standardized benchmarking platform for evaluating robotic ability to understand and manipulate garments effectively, fostering innovation in policy learning, model reasoning, visual perception and representation for deformable object handling.
REAL-I: The 1st Real-world Embodied AI Learning Challenge
The REAL-I Challenge is designed to accelerate progress in embodied intelligence. As robotics moves beyond simulation, the field needs large-scale data, standardized evaluation, and direct access to physical systems. REAL-I brings these pieces together in one place:
Open access to real robots, equipped with unified benchmarking and evaluation tools.
A large-scale industrial dataset, collected with homogeneous dual-arm humanoid robots.
Significant prizes and rewards, designed to recognize outstanding research contributions.
The 11th Robotic Grasping and Manipulation Competition
Since 2016, RGMC engages the community with numerous manipulation tasks in manufacturing, service robots, and logistics areas. This year’s edition focuses on Object Picking in Clutter, Mobile Manipulation, Human Robot Object Transfer, and Cloud Manipulation.
SAI Soccer Challenge: Multi-Agent Reinforcement Learning with Sim2Real Finals on Booster K1
The ICRA 2026 Sim2Real Humanoid Soccer Challenge invites participants to develop intelligent humanoid control policies capable of mastering individual soccer skills, coordinated multi-agent play, and real-world Sim2Real transfer. Teams will train their agents using the SAI Soccer Simulation Suite and deploy them on physical Booster K1 humanoids during the on-site finals.
The competition consists of two stages, each testing different aspects of soccer intelligence:
Team Play Ladder (Simulated 3v3): Multi-agent teamwork under bandwidth-limited communication. Single skill soccer envs emphasizing core abilities such as passing, shooting and defending will also be provided to aid training.
On-site Sim2Real Finals: Deployment of learned policies on real Booster K1 robots during ICRA 2026, including calibration, practice, and knockout matches.
This challenge highlights not only mastery of individual skills but also the ability to generalize across tasks and transfer capabilities from simulation to the physical world.
The 27th RoboRacer Autonomous Racing Competition
Roboracer Autonomous Racing is a semi-regular competition organized by an international community of researchers, engineers, and autonomous systems enthusiasts. The teams participating in 27TH Roboracer Racing Competition at ICRA 2026 will build a 1:10 scaled autonomous race car according to a given specification and write software for it to fulfill the objectives for the competition: Don’t crash and minimize laptime.
The BARN Challenge 2026
The BARN Challenge will take place primarily on the simulated BARN dataset and also physically at the conference venue in Yokohama.
The BARN dataset comprises 300 pre-generated navigation environments, ranging from easy open spaces to difficult highly constrained ones, and an environment generator that can generate novel BARN environments. The task is to navigate a Clearpath Jackal robot from a predefined start to a goal location as quickly as possible without any collision. The Jackal robot will be standardized with a 2D LiDAR, a motor controller with a max speed of 2m/s, and appropriate computational resources. Participants will need to develop navigation systems which consume the standardized LiDAR input, run all computation onboard using the provided resources, and output motion commands to drive the motors. Participants are welcome to use any approaches to tackle the navigation problem, such as using classical sampling-based or optimization-based planners, end-to-end learning, or hybrid approaches. The following infrastructure will be provided by the competition organizers:
- The 300 pre-generated BARN environments
- The BARN environment generator to generate novel environments
- Baseline navigation systems including classical (default DWA, fast DWA, and E-Band), end-to-end learning (e2e RL and LfLH), and hybrid (APPLR) approaches
- A training pipeline running the standardized Jackal robot in Gazebo simulation with Robot Operating System (ROS) Melodic (in Ubuntu 18.04), with the option of being containerized in Docker or Singularity containers for fast and standardized setup and evaluation
- A standardized evaluation pipeline to compete against other navigation systems
What Bimanuals Can Do
In 2026 ICRA WBCD competition, participants are invited to use a wide range of ways: puppeteering, VR, exoskeleton, hand-held grippers, movement-tracking gloves, or algorithmic human hand sensing to perform challenging and practically valuable manipulation tasks and collect data meanwhile. We will evaluate the teams in terms of their task completion quality, data collection speed, and performance of policy learned using the collected data.