Projects
Drone Pollination [Ongoing]
Design and development of active roller based pollinator mechanism to attach on drone to boost paddy pollination.
Collabrative Manipulator [Ongoing]
Design and development of impedance control and force-motion controller for a collaborative manipulator to interact with environment
Aerial Manipulation for Pick and Place
An adaptive sliding mode controller has been used on aerial manipulator to pick and place a payload.
Task space control for aerial manipulator
A task space adaptive controller has been used to push a box lying on table to overcome the state dependent uncertainty and friction forces.
SEA based aerial manipulator
Series Elastic Actuator (SEA) based aerial manipulator is used for shock absorbance during the hammering.
Wall cleaning using aerial manipulator
Demo of aerial manipulator application to clean a glass/wall. The external reaction from the wall is making the drone very unstable and aggressive.
Swing-attenuation for a drone carrying a cable-suspended payload
Our recent work on underactuated control of aerial transportation of a suspended payload via quadrotor demands a controller to simultaneously track the desired path and stabilize the payload swing angles. So we proposed a controller to tackle unknown state-dependent uncertainty without aprior knowledge for aggressive maneuvering and human-interaction during clasping and unclasping of payload.
Real Time Obstacle Avoidance for Quarotors using Fast Planner with Octomapping
Fast-Planner (HKUST) is developed aiming to enable quadrotor fast flight in complex unknown environments. It adopt a kinodynamic path searching method to find a safe, smooth , kinodynamic feasible, and minimum-time initial trajectory in the discretized control space. It uses B-spline optimization, which incorporates gradient information from a Euclidean distance field. A real time harware implemenation is done on custom drone equipped with Intel NUC i7 onboard SBC, Pixhawk flight controller and Realsense D-455 RGBD Camera, where Octomap is build using Point cloud data. The locaization is done using GPS and IMU of Pixhawk. The gazebo simulation is also done in PX4-Autopilot SITL.[Hardware] [Simulation] Github]
Autonmous navigation of quadrotor using a Real-Time Trajectory Replanning
EWOK (TUM) is a local replanning algorithm that can handle unmodeled (possibly dynamic) obstacles while keeping the MAV close to the global trajectory. It store information about the environment around the MAV in an occupancy grid stored in a three-dimensional circular buffer, which moves together with a drone, and represent the trajectories by using uniform B-splines. A real time harware implemenation is done on custom drone equipped with Intel NUC i7 onboard SBC, Pixhawk flight controller and Realsense D-455 RGBD Camera, where local mapping is build using Point cloud data. The locaization is done using GPS and IMU of Pixhawk.[Hardware] [Github]
Autonomous Safe Landing of Quadcopters
It is a local planner to find safe area to land. At the land position, the vehicle will start to descend towards the ground and if the ground is flat, the vehicle will continue landing. Otherwise it will evaluate the close by terrain in a squared spiral pattern until it finds a good enough ground to land on. Autonomous landing system is upgraded with tunable parameters such as; The size of spiral trajectory for searching landing area; A landing threshold to vary according to height of landing gear. It is implemented on quadcopter having a downward facing D455 camera with Intel NUC i7 SBC and Pixhawk flght controller.[Simulation] [Github]
Model Predictive Control for Unicyclic Robot Trajectory Tracking
This work presented an application of MPC to the problem of trajectory tracking of a nonholonomic wheeled mobile robot. The solution of the optimization problem through a standard quadratic programming method using cvxpy package. The obtained control signals were such that the constraints imposed on the control variables were respected. Using successive linearization approach, the optimal control problem was successfully solved arising the possibility of a real time implementation. A matlibplot simulation and Gazebo simulation is done.[Github Matlibplot] [Github Gazebo]
Virtual Structure Fromation Control
Leader Follower Fromation Control
MINION 2.0: A Multi-Robot Payload Transport System
A custom made skid steer mobile robot is build to carry payload inside warehouse where co-operative multi robot will move in a formation. The robot has Arduino as lower level controller and ROS is installed in Pi as SBC. A Leader-follower formation control algorithm has been used for the coordination of multiple agents for maintaining a formation. Robots' formation can shrink and expand according to it's environment. [Github]
Trajectory tracking for six wheeled multi-terrain robot
A simple PID based trajectory tracking implemenatation is done for multi-terrain robot where pre-defined latitude and longitude is tracked using GPS and Magnetometer, while avoiding the obstacles between the waypoints. GPS Data is used to find "Target Heading" and Magnetometer Placed on robot give us "Current Heading" of robot. And difference beteern "Target and Current Heading" will decide angular velocity of robot. The unknown obstacle is detected by ultrasonic sensor.[Github]