Problem
Robotic mining automation needs perception that can find the right object, estimate where it is, and provide a reliable signal before a physical system commits to motion.
Back to systemsConnected object detection to position estimation instead of stopping at bounding boxes. Used simulation to evaluate robotic vision behavior before physical trials. Worked in an R&D setting where perception had to support physical task execution.
ABB Robotics R&D
Robotic Vision for Mining Operations
Robotics R&D Engineer / Apr 2023 - Oct 2023
Built robotic perception workflows for mining automation using ROS 2, YOLO detection, point-cloud position estimation, and Gazebo simulation.
ROS 2YOLOPoint cloud
My Role
Built and simulated robotic vision components that connected detection, point-cloud reasoning, and ROS 2 integration into a workflow suitable for R&D decision-making.
Impact
A robotic vision R&D workflow that turned perception output into spatial information a robotic system can reason about.
System Architecture
- YOLO detection identifies objects of interest.
- Point-cloud processing estimates spatial position for action planning.
- ROS 2 connects perception output to robotic system logic.
- Gazebo simulation validates behavior before real-world hardware risk.
Tools
ROS 2YOLOPoint cloudsGazeboPython
Implementation Highlights
The useful work sits between camera signal and production decision.
Next Step