About This Session
Governments around the globe spend billions of dollars every year to train soldiers to fight wars. However, no nation with an active military force is in a state of perpetual war. This leads to a drop in soldiers’ fitness and combat readiness over a period of time.
Now imagine human soldiers are replaced with humanoid robots that have been trained to pick out enemies and finish them off with precision accuracy. There is no emotion or lethargy in those wires, no fatigue in those motors and there will never be a case of lack of practice.
This talk presents the techniques in Artificial Intelligence that have been developed on a humanoid robot to pick up a gun and practice shooting. The training was done using a NERF gun. In less than 5 tries, the robot achieves an accuracy of greater than 95% on stationery targets. That is a success rate greater than trained marksmen. Future work involves dealing with moving targets, especially those that would do everything to avoid getting hit. The code has been written in C++, with simulations done in MATLAB. The algorithms presented can be easily implemented in all languages on robots with compatible hardware.
Time: 10:45 AM Sunday Room: 4221
The Speaker(s)
unassigned , Microsoft
unassigned