This repository contains homework assignments (HW1 to HW4) for a course on Intelligent Perception and Decision Making. Each homework focuses on different aspects of computer vision, machine learning, robotics, and path planning.
- Utilize
load.pyto collect data andbev.pyfor BEV projection. - Capture screenshots and project points from BEV image to the front image.
- Collect data using
load.pyand reconstruct point clouds withreconstruct.py. - Employ ICP (Iterative Closest Point) algorithm for environmental reconstruction and trajectory creation of ground truth and estimated camera poses.
- Generate data divided into test, train, and annotations.
- Transform data into
.odgtformat and train custom models. - Collect RGB, depth, and annotation data in the Habitat environment.
- Produce predicted semantic segmentation and colorize ground truth and estimated annotations.
- Reconstruct the environment using semantic segmentation.
- Obtain BEV images of the entire environment.
- Plan paths using the RRT (Rapidly-exploring Random Tree) algorithm.
- Display paths in the Habitat environment and save the path planning results.
- Implement and test forward kinematic algorithms and Jacobian matrices for each pose.
- Implement and test the inverse kinematic algorithm.
- Complete manipulation pipeline including pose matching, robot movement control using inverse kinematics, and motion planning with RRT-Connect for collision-free paths.
This README provides an overview of the assignments and their objectives in the field of intelligent perception and decision making. Each link directs to the detailed README of the respective homework for further information.