In October 2018, I got selected as the part of the IIT Kharagpur contingent to participate in the 7th Inter-IIT Tech meet, IIT Bombay. I was a part of the team working for the Tata Centre for Technology and Design(TCTD) Challenge. The problem statement of the event was to provide a solution that reduces the drudgery of farmers by providing intelligent automation. Thus after a lot of brainstorming and interviewing the nearby farmers, we found out that farmers face the most difficulty in the task of fruit plucking and seed sowing. We came up with a novel solution that could cost-effectively solve this problem. We designed and prototyped an autonomous robot capable of performing the following task-
- Fruit detection and plucking
- Seed sowing
- Weedice spraying
Is was a part of the planning, controls and embedded team. My major contribution was towards the overall electronic architecture of the robot along with the localisation of the robot. Extended Kalman Filter was used for the same. Various sensors like IMU and rotary encoders were used for the purpose of localisation. The whole robot would only work on a pre-mapped environment and a pre-generated path. Thus it was necessary to implement a path-tracking algorithm which would help the bot run on the pre-determined path precisely. I implemented the Pure-pursuit algorithm for the same. Since the robot was a differential drive based, the longitudinal velocity and the angular velocity were then converted into independent left and right motor velocities.
The major task was the controls of the 4-DOF robotic manipulator. A servo motor along with an IMU for extra feedback was used to control the yaw angle. Two linear actuators were used to control the pitch angle and the linear extension of the arm. Sharp IR sensors were used to provide feedback of the linear actuator extension. The end-effector had a camera which was used for fruit detection. In addition to that, it had a pressure sensor calibrated for different types of fruits. The opening and closing of the gripper were controlled by three synchronised servo motors. Each of the DOF was controlled independently. A PID controller was used to achieve the target state which was solved by using inverse kinematics on the co-ordinates of the fruit given by the software module.
Demonstration of Seed-Sowing