Robocon 2022: Lagori (同创辉煌)
Published:
Award: National Runner-up (First Prize)
ROBOCON 2022 China National Contest
Project Details
This project involved designing two logistic robots (R1 and R2) to autonomously perform tasks including ball recognition, grasping, and throwing, as well as cylinder identification and stacking in a complex environment.
My Role
I was primarily responsible for the visual perception algorithms for both robots.
Object Detection (YOLO + ONNX): Trained a YOLO neural network for ball detection. To ensure real-time performance in C++, the model was converted to ONNX format and deployed locally using the ONNX Runtime.
Point Cloud Processing (PCL): Utilized the Point Cloud Library (PCL) to process data from depth cameras. By implementing segmentation and clustering algorithms, I successfully solved for the pose, size, and relative position of cylinders in complex environments, guiding the robotic arm for grasping.
