A real-world robotics challenge with a $180K prize pool, where innovation and industry impact collide.
We’re standing at an inflection point in robotics: electronics assembly, especially dexterous manipulation remains one of the biggest open problems in industry today. Tasks like handling flexible cables or inserting connectors during electronics assembly, are still exceedingly hard for robots due to the complexities of perception, planning, and control. These challenges directly limit automation in factories and supply chains worldwide.
Intrinsic and Open Robotics proudly present theThe AI for Industry Challenge. Developed in collaboration with Nvidia and Google DeepMind, the program is designed to transform the industrial landscape.It’s an open challenge inviting developers, engineers, roboticists, and research teams to apply the latest in AI, simulation, and robotic control to a real-world dexterous task that has challenged both academia and industry for years. Participants in the AI for Industry Challenge will help innovate and push forward the solution for intelligent robotic manufacturing.
What’s Really Exciting About This Challenge?
- This isn’t a toy problem. Dexterous cable manipulation mirrors real industrial problems: think endless manual wiring in electronics assembly that could be automated with the right AI and perception breakthroughs.
- Open tools, open creativity. Use open-source simulators and robotics stacks of your choice. This is an opportunity to push your own methods – whether it’s reinforcement learning, control-theoretic approaches, or novel computer vision pipelines.
- Go from simulation to reality. Finalists will run their solutions on physical hardware, a rare chance to test in a real industrial scenario.
- Create or join a team and work together. While solo entries are welcome, teams (up to 10 members) allow you to combine strengths in perception, learning, control, and system design for the best shot at success.
What Participants Can Expect
Participant teams will:
- Train models to manipulate and insert electronic cables using simulation tools (e.g., Gazebo, Isaac Sim, MuJoCo) and ROS interfaces.
- Build complete solutions using Intrinsic Flowstate’s development environment and the Intrinsic Vision Model (IVM). Note: Top 30 teams only.
- Deploy on real robots on Intrinsic’s physical workcell during the final phase, bridging the sim-to-real gap with your own algorithms. Note: Top 10 teams only.
If that isn’t enough to get you ready to accept the challenge today, here are the prizes…
The Prizes & Challenge Timeline
There are 4 phases of the challenge. Timelines for each phase will be announced soon.
- Qualification phase: All participants can utilize Intrinsic’s open source platforms, or any platform of their choosing to train a model and tackle the challenge’s first assembly tasks.
- Phase 1: The 30 top teams will be given access to the Intrinsic Flowstate developer environment and one month to solve the end-to-end assembly tasks.
- Phase 2: The 10 finalist teams will be able to deploy their simulation onto a physical robot cell at Intrinsic HQ and evaluate their model on industrial-grade hardware.
- Overall winners: 5 individuals/teams will be announced to share a cash prize worth up to $180,000.
Here’s how the $180,000 prize pool breaks down:
- 1st Place – $100,000
- 2nd Place – $40,000
- 3rd Place – $20,000
- 4th & 5th Place – $10,000 each
Registration is open until April 17, 2026, and the challenge officially kicks off March 2, 2026 with a toolkit release to get you started. Sign up on the official website to keep up to date on releases and news.
Who Should Enter?
- Computer vision pros & researchers who want to apply perception to hard manipulation tasks
- Machine learning experts researching policies, sim-to-real methods, reinforcement learning
- Robotics engineers who know ROS, Gazebo, MuJoCo, Isaac Sim or other advanced simulators
- Cross-disciplinary teams combining perception, planning, and control expertise
This is a key moment in robotics – the community’s collective progress is converging with the tools and infrastructure needed to make real industry-wide impact.
How to Get Started
Visit the official challenge page to register and join a team, or compete as a team of one before the April 17 deadline.
Every major breakthrough in automation has started with a bold challenge – this is yours. Build, innovate, compete, and help solve one of the most meaningful problems in modern robotics. Prizes and bragging rights are up for grabs. Let’s go!
The post Calling Roboticists & Vision Experts: Tackle Dexterous Manipulation and Win Big in the AI for Industry Challenge appeared first on OpenCV.