Our build and extension of the Stanford Pupper Robot. Codename: C.E.R.B.A.R.I.S.
Introduction | Team | Hardware | Software Overview | Software Setup | Computer Vision | Collision Avoidance | Web Interface | Odometry | Behavioral Control |
For this project we chose to use 3 front mounted IR object sensors. These sensors switch from HIGH to LOW when an object is detected. They can detect objects up to 30cm though some objects do not reflect IR well and thus have a much shorter detection range (such as black plastic cabinets). The detection range can be tuned by a screw on the back of each sensors.
The hardware communcation is achieved through the pigpio package. For ease of use the pupperpy.object_detection.ObjectSensors object can be instatiated to quickly and easily grad sensor data.
from pupperpy.object_detection import ObjectSensors
sensors = ObjectSensors()
sensors.read() # returns dict with keys 'left', 'right' and 'center' which have
# boolean values
The current Control flow uses this information to avoid collisions by simply turning away from any detected objects. If the object is only on the left or right sensor then the robot continues moving forward while turning, but if the object is also present on the center sensor – or only present on the center sensors— then the robot stops and turns until the object is no longer detected.
Going forward object avoidance can be made much smarter. The simplest and probably most useful improvement would be to turn back after dodging an object that only appeared in only the left or right sensor, this way it would acutally go around an object and not just turn and walk off in an awkward direction, possibly completely losing track of a previously tracked target. Additionally, heading information could be used from the IMU to help relocalize a target that was lost by a suddenly appearing obstacle.
One weakness of the current sensor array is low obstacles that could trip the robot – such as the legs of rolly chairs. It may be better to mount the sensors pointing at a slight downward angle so as to detect and avoid those.