Atharva Bhorpe

Robotics master’s student at RWTH Aachen

I work on robot learning, control, and simulation. My background is in mechanical engineering and industrial automation.

Projects

Learning / Navigation

Recorded tutorial: training metrics and predicted and ground-truth paths on an occupancy grid.

What I built: I built a recursive transformer for occupancy-grid path planning.
My contribution: I evaluated path success, route optimality and latency against A* and parameter-matched CNNs.
Tools: PyTorch · Transformers
Scope: Approximately 1.2M parameters. Grid evaluations, not physical-robot tests.

Hardware / ROS

What I built: Our five-student team built an affordable ROS mobile-robot prototype.
My contribution: I integrated the ROS stack with Arduino firmware on an ESP32.
Tools: ROS · Arduino · ESP32 · Fusion 360
Scope: Laptop-based student prototype for robotics education and research.

Simulation / Manipulation

Learning in simulationUR5e · downward SFP insertion · fixed ports.
Illustrative motion, not policy execution.

What I built: Our team trained imitation-learning policies for connector insertion with a simulated UR5e arm.
My contribution: I focused on policy tuning and data collection.
Tools: LeRobot · PyTorch · Diffusion Policy · Gazebo
Scope: Gazebo simulation with SFP and SC fiber-optic plugs.

Sensing / Control

From sensor input to robot motionMPU6050: X/Y · encoder: Z · IR: gripper.
Illustrative motion · simulated seven-axis Gen3.

What I built: I built a sensor-based controller for a simulated Kinova Gen3 arm.
My contribution: I connected IMU, rotary encoder and IR inputs to ROS 2 and MoveIt through an ESP32.
Tools: ROS 2 Humble · MoveIt · ESP32 · Arduino
Scope: Sensor-to-simulation control over WiFi and micro-ROS.