Simulation Environment: Built simulation environment in PyBullet with a customized map and a PR2 robot.

Kalman Filter Implementation: Implemented Kalman Filter to achieve localization by continuously updating its estimation of a system's state using a combination of motion model and incoming sensor data.

Particle Filter Implementation: Implemented Particle Filter to achieve localization by using particles to represent system states, updating and resampling them with new sensor data to identify the most probable state.

Network Structure Revision: Revised and tested MonoCon on the KITTI benchmark, specifically for car detection in various scenarios and evaluated the performance based on 3D bounding box predictions.

Dataset and Training: Data augmentation techniques were used to enhance the model's robustness. Extensive experimentation was conducted to optimize hyperparameters, ensuring improved model performance.

Differential Dynamic Programming (DDP) Implementation: Utilized DDP to implement a trajectory optimization algorithm for inverted double pendulum on a cart task.

Model Predictive Path Integral Control (MPPI) Implementation: Implemented the MPPI algorithm to achieve the same task goal.

Cost Function Design: Developed appropriate cost functions for both methods.

Position Calibration: Executed precise calibration and transformation from the depth camera coordinate to the world coordinate.

State Machine Control: Devised a sophisticated state machine to control the arm's movement in the workplace systematically.

Block Detector: Detected blocks of different colors and determined object grabbing and placement positions.

Kinematic Algorithm: Conducted thorough debugging of the inverse kinematic code, ensuring the arm's movement aligned with expectations.

Real-time Positioning System: Utilized 2D Lidar to construct a real-time positioning system for the robot, incorporating a Monte Carlo (MCL) positioning algorithm based on odometer and laser ranging, allowing real-time location on known maps.

Path Planning: Developed A* algorithm enabling the robot to plan the optimal path within the environment.

Automatic Exploration and Obstacle Avoidance: Integrated a robotic system with automatic exploration capabilities in unknown areas and obstacle avoidance features for safe and efficient task completion.

Environment Perception and Obstacle Avoidance: Utilized Simultaneous Localization and Mapping (SLAM) with 2D Lidar for real-time map creation and obstacle avoidance.

Positioning and Navigation: Implemented the adaptive Monte Carlo positioning (AMCL) algorithm for autonomous navigation.

Path Planning: Employed the rapidly exploring random tree (RRT) algorithm for efficient path planning.

Object Pick-up: Integrated a monocular camera for object positioning, with a 4-DOF manipulator controlled by an inverse kinematics (IK) algorithm for precise target grabbing.

Electrical System Design: Diagnosed and resolved issues in the existing electrical system, constructing a new and enhanced system for the vehicle.

Battery Bank Development: Designed and built a new battery pack, significantly enhancing the vehicle's power.

Motion Planning: Specialized in motion planning, employing MoveIt! for robot arm trajectory and integrating Dijkstra's algorithm with Dynamic Window Approach (DWA) planner for base path planning.

Platform Interface: Developed Python scripts for low-level hardware control and interfaced with complex simulation environments within the ROS package, achieving integration and manipulation of virtual environments for real robot testing and development.

Module Integration: Integrated motion planning with SLAM and perception to form a complete autonomous robot system.

Simulation and Track Design: Developed a precise simulation environment tailored for a custom racetrack.

Advanced Control System Implementation: Upgraded the control system from conventional PID to the sophisticated Model Predictive Control (MPC) algorithm.

Collision Avoidance: Introduced a competition-oriented training strategy using MPC agent as a rival vehicle.