ICRA 2025 Competitions
The BARN Challenge 2025
Designing autonomous robot navigation systems has been a topic of interest to the robotics community for decades. Indeed, many existing navigation systems allow robots to move from one point to another in a collision-free manner, which may create the impression that navigation is a solved problem.
However, autonomous mobile robots still struggle in many scenarios, e.g., colliding with or getting stuck in novel and tightly constrained spaces. These troublesome scenarios have many real-world implications, such as poor navigation performance in adversarial search and rescue environments, in naturally cluttered daily households, and in congested social spaces such as classrooms, offices, and cafeterias.
Meeting these real-world challenges requires systems that can both successfully and efficiently navigate the environment with confidence, posing fundamental challenges to current autonomous systems, from perception to control. Therefore, we propose The Benchmark Autonomous Robot Navigation (BARN) Challenge as a benchmark for state-of-the-art navigation systems and to push the boundaries of their performance in these challenging and highly constrained environments.
We have successfully organized the 1st, 2nd, and 3rd BARN Challenge in Philadelphia, London, and Yokohama respectively (see the award ceremonies in the first three images above). In the 1st BARN Challenge, five international teams from three countries participated in the simulation competition, three of which were invited to compete with each other in the physical obstacle courses set up in Philadelphia. The 2nd BARN Challenge has doubled its size with ten international teams from six countries participated in the simulation competition, six of which were invited to compete in London. The 3rd BARN Challenge has six teams from five countries in the simulation competition, four of which were invited to compete in Yokohama. This is the 4th BARN Challenge.
Robotic Grasping and Manipulation Competition
The 10th Robotic Grasping and Manipulation Competition (RGMC) at ICRA 2025. RGMC has been running for nine editions at either the IEEE/RAS International Conference on Robotics and Automation (ICRA) or the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), since 2016 [1]. Across the editions, we engaged the community with various tasks for manufacturing, service robots, and logistics. These tasks include assembling and disassembling boards, hand-in-hand grasping, picking and placing various objects, pouring liquids into a cup, bin picking, rearranging and setting formal tables, folding and unfolding cloths, and receiving objects handed over by a person. RGMC will advance research and technology towards more realistic scenarios that can be encountered in daily activities at home or in warehouses, and will increase the interest of the community in solving the open challenges associated with robotic grasping and manipulation.
The 4th Robotic Sim2Real Challenges
Simulators have significantly accelerated engineering development in robotics, making a profound impact on advancing technological progress in this field. However, as simulation models cannot perfectly replicate robots’ behavior in real world, there exists a notable sim-to-real gap.
Given the high cost of acquiring large amounts of real-world robot data, data augmentation using simulators is an essential step towards advanced robot intelligence. Robotic Sim2Real Challenges aim to encourage more students and researchers to explore the cooperative method using simulators and real-robots, and to systematically investigate effective approaches for their matching.
Hopefully, this will lay a theoretical foundation and provide practical experience for the sustainable development of embodied intelligence.
Robotic Sim2Real Challenges have already been held for 3 times in conjunction with ICRA Since 20221. For ICRA 2025, we propose the 4th Robotic Sim2Real Challenges with two tracks: the mobile manipulation track and the flying robot track. The former is a consecutive event of
pervious AgileX Sim2Real Challenge (AXS). The latter is a consecutive event of pervious RoboMaster University Sim2Real Challenge (RMUS) with a new flying robot instead of DJI’s RoboMaster EP (DJI’s product EP has been discontinued.). For both tracks, we provide a photorealistic simulator supporting end-to-end embodied AI model training and parallel
reinforcement learning training.
AI Olympics With RealAIGym: Towards Global Policies for Swing up in the Real World
As artificial intelligence gains new capabilities, it becomes important to evaluate it on real-world tasks. While software such as ChatGPT has recently revolutionized certain areas of AI, athletic intelligence seems to still be elusive in the AI community. To have better robots in the future which can perform a wide variety of dynamic tasks in uncertain environments, the physical or athletic intelligence of robots must be improved. However, this is quite challenging.
In particular, the fields of robotics and reinforcement learning (RL) lack standardized benchmarking tasks on real hardware. To facilitate reproducibility and stimulate algorithmic advancements, the 3rd AI Olympics competition is being proposed to be held at ICRA 2025 in Atlanta following the inaugural run at IJCAI 2023 in Macau , click here for a video summary), based on the RealAIGym project and 2nd edition of this competition at IROS 2024 in Abu Dhabi. While the inaugural run of this competition focused on comparing the performance of various learning and control methods on the real acrobot/pendubot systems, the second edition at IROS focused not only on achieving the best scores but also on achieving robustness to disturbances during the execution.
At ICRA 2025, the focus of the challenge would be to develop a global policy that can solve the swing up problem from any configuration in the state space thus improving the controllers in the 3rd edition of this challenge. This will be tested by randomly disturbing the system with very large (compared to IROS 2024) external disturbances while it performs the swingup task. The challenge will involve two stages: simulation and realrobot experiments where teams (and their agents) can compete to get the highest score to win some cool prizes! We invite people from all communities (AI/ML/RL, Optimal Control, Heuristics, etc.) to participate in this competition on a set of standardized dynamic tasks on well-known prototypical systems using standardized hardware.
PhyRC: Physical Robotic Caregiving Challenge
What Bimanual Teleoperation and Learning from Demonstration Can Do (WBCD)
What Bimanual Teleoperation and Learning from Demonstration Can Do (WBCD) Today (in 2025) There have been lots of awesome works around the topic of bimanual teleoperation and learning from demonstration, employing various human sensing systems and robot embodiments. To name a few, ALOHA, Mobile ALOHA, DexCap, Open-TeleVision, HATO, GELLO, AirExo, UMI, etc.
However, researchers design and evaluate on tasks that they come up with and perform teleoperation themselves. Also, there are lots of metrics that are critical for industry that are not evaluated in the academic research works, such as motion speed, system robustness, cost efficiencies, and difficulty to learn an effective policy.
Therefore, there has been a demand in the industry to set up a common set of tasks benchmark that is practically valuable and reflects various axes and levels of difficulty. The goal of this competition is to bring together researchers with industry and figure out the best fits between
valuable tasks and technology solutions.
In this competition, participants are invited to use a wide range of ways: puppeteering, VR, exoskeleton, hand-held grippers, movement-tracking gloves, or algorithmic human handsensing to perform challenging and practically valuable manipulation tasks and collect datameanwhile. We will evaluate the teams in terms of their task completion quality, data collection
speed, and performance of policy learned using the collected data.
Earth Rover Challenge 2
With the rise of generalist foundation models for robot navigation, new frontiers involving challenging open-world and open-vocabulary mobility scenarios are now of immense interest to the robotics community.
Our competition aims to explore the possibility of globally distributed in-the-wild navigation evaluations at an unprecedented scale, along with the release of a substantial real-world navigation dataset collected from 10+ cities (>5k hours).
Leveraging a large fleet of outdoor navigation robots deployed across multiple cities, we aim to study whether state of the art open-world autonomous navigation models are able to effectively operate in truly open-world settings and how they fare against human teleoperated performance under the same environments.
We propose a distributed competition across remote environment scenarios spanning at least 4 separate cities (eg. Atlanta, Madrid, Taipei, Stockholm, etc.), where competition participants need to deploy their policies into realistic goal-oriented navigation scenarios, potentially without known maps. This competition will test the robustness, generalization, and safety of navigation capabilities of robot foundation models.
Quadruped Robot Challenges(QRC)
Autonomy requires a good mobile platform, perception technologies, navigation technologies, etc.
Quadruped robots were never tested in a realistic common challenging environment in an autonomous way before ICRA 2023. Quadruped robots are inherently dynamic; thus, their competition is interesting to the audience and public and may draw the attention of potential sponsors.
Quadruped Robot Challenges (QRC) was inaugurated in ICRA 2023 to test teams for autonomous traversability on various kinds of terrains. At Stage 1, each robot runs the test field alone, solo start.
Due to the short notice of the game, many teams used remote control. Only two teams challenged in autonomy. In Stage 1+ which is about the mobility with inspection test at ICRA 2024 and IROS 2024, we witnessed more teams use autonomy and an exploration mission is still difficult for autonomous robots.
As a next step, ICRA 2025 QRC will move to Stage 2 where the multiple robot collaboration concept will be introduced. Teams may need to deal with more advanced tasks, dynamic obstacles, real-time perception, and planning in this stage.
The 24th Roboracer Autonomous Grand Prix Competition
2025 Theme: MAD-GAMES: Multi-Agent Dynamic Games
Roboracer Autonomous Racing https://roboracer.ai/is an international competition organized by a community of researchers, engineers, and autonomous systems enthusiasts. Roboracer is used in over 89 universities for research, teaching and participation in competitions. The teams
participating in this Roboracer Grand Prix at ICRA 2025 will build a 1:10 scaled autonomous race car according to a reference specification. The teams write software for this scaled race car to fulfill the objectives for the competition: Don’t crash, minimize lap time and overtake all opponents. This competition considers the challenging scenario of multi-agent (many-on-many) racing and aims to assess the state-of-the-art for perception, planning and competitive control algorithms in real world scenarios, inviting contributions from a broad spectrum of research directions spanning model-based classical planning and control to model-free reinforcement learning end-to-end agile navigation approaches.
Following the success of the Roboracer Autonomous Racing competitions at ICRA’22 Philadelphia (100+ participants and 20 teams), and ICRA’23 London (130+ participants and 24 teams), ICRA’24 Japan (135+ participants, 19 teams) for the ICRA 2025 competition we propose to set up the 24th Roboracer Autonomous Grand Prix in the form of both: (A) In-person hardware-based competition and a (B) virtual simulation-based race. These versions of the competition allow for accessible and equitable participation of teams worldwide.
