Research Scientist

• Gemini Robotics, focusing on intelligent whole-body control.

Research Intern

• Proposed and ran research on data-driven reinforcement learning.
• Set up reinforcement-learning training infrastructure on a cluster.

Researcher

• Researched AI for robots (deep learning, reinforcement learning).
• Aimed at robots for decommissioning the Fukushima nuclear power plant.

Intern

• Worked on a lunar exploration rover.
• ToF camera data processing and elevation mapping.
• Built a Gazebo simulator package for a rover on the lunar surface.

Part-time

• Built a drone with a custom flight controller and Jetson TX2.
• Autonomous flight with visual-inertial odometry, model predictive control and local obstacle-avoidance planning on a 3D signed distance field.

Ph.D.

• Thesis: “Bridging Perception and Control for Legged Locomotion and Navigation in the Wild” (defended April 17, 2025).
• Reinforcement learning for perceptive locomotion of the quadrupedal robot ANYmal.
• Team CERBERUS member at the DARPA Subterranean Challenge, which won first prize.

Exchange student

• Mapping and trajectory planning for legged robots at the Robotic Systems Lab.
• Multi-agent time-based decision-making for the search-and-action problem, for the MBZIRC challenge at the Autonomous Systems Lab.

M.S. in Aeronautics and Astronautics

• UAV/UGV autonomous cooperation: a drone attaches a tether so a rover can climb a cliff by winding it.
• GPA 3.92/4.0.

B.S. in Aeronautics and Astronautics

• Concentration in astronautics, control engineering and artificial intelligence.
• Thesis: self-localization with active attachment of AR markers by an autonomous UAV.
• Major GPA 3.78/4.0, overall GPA 3.59/4.0.

IEEE ICRA Best Paper Award in Service Robotics

Learning to walk in confined spaces using 3D representation

2nd prize, SaTML 2024 LLM Capture-the-Flag

Team RSLLM · $1,000 prize and travel grant

1st place, DARPA Subterranean Challenge Final

Team CERBERUS · $2,000,000 prize

Machine learning system, domain conversion device, and machine learning method ↗

R. Sakai, N. Kimura, T. Miki · US 2020/0184269 A1