Research Taste
Policies that can leave simulation
I care about learned motor skills, sim-to-real transfer, and robot systems where perception, action, dynamics, and hardware constraints are designed together.
Robot learning engineer · Trondheim
Building learned robot control systems that survive contact with hardware.
I am Daniel Rosmæl Skauge. I work where deep reinforcement learning, real-time control, ROS 2, and hardware integration meet, with a bias toward getting policies out of simulation and onto physical robots.
Strongest Proof
Deep RL for legged robotics
Master's thesis graded A on a jumping quadruped robot in Mars gravity, using deep reinforcement learning for dynamic locomotion under unusual physical constraints.
Current Work
Humanoid control infrastructure
Building close to the robot at InMind: C++ control software, ROS 2 systems, VR teleoperation, calibration, diagnostics, and hardware integration.
Research Engineering Fit
- Deep reinforcement learning
- Learned control policies
- Sim-to-real transfer
- Robot learning systems
- ROS 2 and real-time Linux
- C++, Python, and PyTorch
- State estimation and calibration
- Hardware debugging