(M/W) AI

Il y a 2 semaines

Paris, Ile-de-France Wandercraft Temps plein


Job Description
Wandercraft is strengthening its Manipulation team to tackle challenging industrial use cases with Calvin, our humanoid robot. We are looking for an intern to explore how simulation can help us train manipulation policies while reducing the amount of data that needs to be collected on the real robot. The internship will combine robotics simulation, imitation learning and hands-on experiments on real hardware, with a strong focus on using simulation to improve real-world data efficiency. During the internship, you will work closely with the Manipulation team to:
• Reproduce our current experimental manipulation setup in a robotics simulator such as NVIDIA Isaac Sim, including the robot, manipulated objects, and relevant elements of the environment.
• Build a pipeline to generate manipulation demonstrations in simulation and use them to train imitation-learning policies.
• Investigate and implement techniques to reduce the sim-to-real gap, for example through domain randomization, system identification, sensor and actuator modeling, or visual augmentation.
• Evaluate the resulting policies both in simulation and on the real robot and analyze which aspects of the simulation are most critical for successful transfer.
• Depending on the progress of the internship, explore Reinforcement Learning approaches to improve existing policies, using simulation to efficiently generate additional experience. Preferred Experience
• Diploma: You are currently pursuing an engineering degree, Master’s degree, or equivalent, with a specialization in robotics, artificial intelligence, control, or a related field. You have a strong interest in robotic manipulation and learning-based robotics, and you enjoy working at the intersection of algorithms and real-world experiments. Experience with some of the following would be particularly valuable:
• Robotics simulation, ideally Isaac Sim
• Machine Learning and PyTorch
• Imitation Learning and/or Reinforcement Learning
• Robotics fundamentals: kinematics, dynamics and coordinate transformations
• Computer vision and image processing
• Python and software development in a Linux environment You are rigorous, and can clearly explain your calculations and algorithms. You are curious and analytical, able to stay up to date with new technologies and learn effectively from online documentation, interaction with the team and literature reviews.
• Required

skills:
• Python
- Autonomous level
• C++
- Basic knowledge
• Machine learning
- Autonomous level
• Imitation learning and diffusion models
- Autonomous level
• Robotics and control systems
- Autonomous level
• Linear algebra, matrices computations, numerical optimization
- Autonomous level
• Computer vision
- Basic knowledge
• Software engineering best practices, collaborative software design, continuous integration, automated testing
- Autonomous level Who are they? At Wandercraft we pioneer intelligent robotics to help people do what they can’t do. We apply the most advanced robotics and AI to expand what’s possible for humans in the fields of healthcare and now industry. Founded in 2012, Wandercraft is globally recognized for creating the world’s first self-balancing walking exoskeleton, Atalante X, which is used in more than 100 rehabilitation centers across four continents helping individuals with disabilities re-learn how to walk, with over a million steps taken each month. Building on over a decade of real-world experience and a proprietary neural network trained on billions of steps, we’re bringing mobility beyond clinical settings and into the home. Our personal exoskeleton, Eve, is a first-of-its-kind device designed to assist wheelchair users in standing and walking in everyday life. Now we’re expanding our platform to support industrial and assistive tasks with the Calvin family of humanoid robots. Built in just 40 days, Calvin-40 is designed to relieve factory workers from hazardous labor. Future versions of Calvin will assist persons with reduced autonomy in their daily activities. Our growth into home and industrial robotics opens new horizons while reinforcing our ability to scale innovation, improve reliability, and accelerate our impact, across our entire portfolio of products and industries. From Paris and New York, our team of over 110 engineers, clinicians, researchers, and operators is united by one mission: to design robotics that improve human life, wherever people need them most We’re not just hiring people: we’re building a team of pioneers, curious minds and problem solvers who want to change the world, step by step. Join us to shape the next generation of medical and humanoid robotics powered by AI and purpose.