This is our submission for event named Hard-Wired conducted by Robotics Club - IITBHU.

There are specific positions for every number on the die and the goal of the game is to rotate the arm by 60⁰ and close the gripper in the least number of turns. You can control the position of the robot arm with hand gestures. We use 3 gestures

Thumbs Up: Moves the arm to a specific position according to number on dice

Thumbs Down: Moves the arm to initial position

Fist : Closes the gripper

1. Hand Gestures
2. Roll a Die Score of the game is (1/number of times the dice rolled) Give a hand gesture after the dice is rolled.

● Gesture Recognition: We removed background from the hand gesture image to get a segmented image and developed a CNN architecture which takes segmented image and classifies hand gestures.

● Dice Number Recognition: We Used blog detection of OpenCV library to detect the number of circles present on the top face of the die.

Run below commands in 3 seperate terminals

roslaunch control run.launch

roscd control
python3 dice.py

roscd control
python3 gesture.py

Gujulla Leel Srini Rohan

Sai Harsha

Ayush Ishan