Intern (m/f/d) Robot Learning

Agile Robots SE · Munich

InternshipGerman needed: none

Still open on 1 Nov

You apply on the company's own website.

What you will do

Agentic Planning: Build and test systems that use language or vision-language models to plan, sequence, and monitor robot tasks over multiple steps.
Sim-to-Real Transfer: Apply domain randomization and realism modeling to reduce the gap between simulated and real sensor and actuator data.
Model Training: Train, fine-tune, or adapt machine-learning models for robot perception, manipulation, or task planning, including adapting an existing model to a specific robot and task.
Model Evaluation: Design and run experiments in simulation and on real robots to measure where a model works and where it fails, and trace failures to the data, the model, or the setup.
Data Preparation: Collect, clean, and annotate robot data, including teleoperation demonstrations and experiment records from real hardware.

Degree Programme: Enrolled in a Master’s programme in Computer Science, Robotics, Machine Learning, Electrical Engineering, or a related field.
Machine Learning: Hands-on experience training and evaluating neural networks in PyTorch or a comparable framework, gained through coursework, projects, or research.
Project Experience: At least one machine-learning project in which you prepared the data and designed the evaluation yourself.
Python Programming: Working proficiency in Python for machine-learning and robotics development.
Robot Simulation: Working familiarity with at least one physics simulator, such as MuJoCo, Isaac Sim, or Gazebo.
Language Agents: Experience building a system in which a language or vision-language model plans or triggers actions in a simulator or on a robot.
Robot Learning: Experience with imitation learning, reinforcement learning, or vision-language-action models.
Sensor Modeling: Exposure to sensor noise models, actuator dynamics, or domain randomization techniques.
Robot Hardware: Experience running code on a physical robot or sensor setup, for example in a university lab or a student robotics team.
ROS 2: Familiarity with ROS 2 and how robot software components communicate.

Agentic Planning: Build and test systems that use language or vision-language models to plan, sequence, and monitor robot tasks over multiple steps.
Sim-to-Real Transfer: Apply domain randomization and realism modeling to reduce the gap between simulated and real sensor and actuator data.
Model Training: Train, fine-tune, or adapt machine-learning models for robot perception, manipulation, or task planning, including adapting an existing model to a specific robot and task.
Model Evaluation: Design and run experiments in simulation and on real robots to measure where a model works and where it fails, and trace failures to the data, the model, or the setup.
Data Preparation: Collect, clean, and annotate robot data, including teleoperation demonstrations and experiment records from real hardware.

Degree Programme: Enrolled in a Master’s programme in Computer Science, Robotics, Machine Learning, Electrical Engineering, or a related field.
Machine Learning: Hands-on experience training and evaluating neural networks in PyTorch or a comparable framework, gained through coursework, projects, or research.
Project Experience: At least one machine-learning project in which you prepared the data and designed the evaluation yourself.
Python Programming: Working proficiency in Python for machine-learning and robotics development.
Robot Simulation: Working familiarity with at least one physics simulator, such as MuJoCo, Isaac Sim, or Gazebo.
Language Agents: Experience building a system in which a language or vision-language model plans or triggers actions in a simulator or on a robot.
Robot Learning: Experience with imitation learning, reinforcement learning, or vision-language-action models.
Sensor Modeling: Exposure to sensor noise models, actuator dynamics, or domain randomization techniques.
Robot Hardware: Experience running code on a physical robot or sensor setup, for example in a university lab or a student robotics team.
ROS 2: Familiarity with ROS 2 and how robot software components communicate.