This work not only addresses immediate challenges but also paves the way for a future where robots seamlessly augment human activities. This project advances robotic capabilities through three key phases. It begins with detailed modelling and controller tuning, ensuring precise kinematic performance. Next, it pioneers motion planning algorithms for agile navigation in complex environments. Finally, it showcases the practical application of robotic manipulation, hinting at their potential to revolutionize various industries.

Model

In the first part of this project, the critical aspects of robot modelling are focused on, which is paramount in robot design and control. The project encompasses a series of tasks that delve into Forward, Inverse, and Differential Kinematics, alongside Robot Control, utilizing a proportional-integral-derivative controller. The robot model consists of five links, including the base link and end effector, connected by three revolute joints.

The project successfully computed the initial D-H table of the robot arm, completed the code for the D-H matrix, and calculated forward kinematics to determine the pose of the robot's end effector. Additionally, it verified points within the robot's workspace and calculated inverse kinematics to ascertain the necessary joint angles for reaching desired positions. The Jacobian was computed for differential kinematics, and controller gains were tuned for optimal robot arm performance, adapting to different scenarios such as increased end effector mass. This project not only demonstrated the robot's ability to follow a desired trajectory with precision but also highlighted the importance of tuning controller gains to adapt to changes in dynamics, paving the way for future advancements in robotic control systems.

Plan

In the second part of this project, advanced motion planning algorithms for Robot DE NIRO, a dual-arm robot mounted on a mobile wheelbase are developed. This explores three key implementations: manual waypoint navigation, the potential field algorithm, and the probabilistic roadmap (PRM) algorithm. The manual approach involved strategically placing waypoints to navigate around obstacles, resulting in a simple yet non-optimal path.

The potential field algorithm treated the robot as a charged particle, using attractive and repulsive forces to navigate, which, while not yielding optimal paths, allowed for reactive adjustments to dynamic environments. The PRM algorithm, on the other hand, utilized random sampling and graph creation to find near-optimal paths efficiently. The project's outcomes demonstrated the effectiveness of these algorithms in enabling the robot to intelligently manoeuvre through complex environments, highlighting the potential for their application in real-world robotic systems. The findings underscore the importance of tailored motion planning strategies in enhancing the autonomy and functionality of robotic platforms.

Interact

In the final part of this project the realm of robotic manipulation and interaction within a virtual environment is comprehensively explored. The project delves into the intricacies of position, velocity, and torque control, utilizing a simplified version of the Natural Interaction Robot (DE NIRO) with dual arms, each boasting seven degrees of freedom. The study meticulously examines the calculation of end-effector orientation using Euler angles, the execution of pick-and-place tasks, and the intricate computations involved in twist and joint velocities.

It further investigates the utilization of null space projection to perform secondary tasks without hindering primary objectives, showcasing the robot's ability to handle complex interactions with its surroundings. The project's outcomes demonstrate the robot's proficiency in performing tasks such as demolishing walls and cleaning tables, highlighting the potential for robots to significantly contribute to various industries through advanced interaction capabilities.