Landing : Athabascau University
  • Blogs
  • The Robotics Primer: Chapter 10

The Robotics Primer: Chapter 10

Open loop control

Using open loop control may be beneficial because it does not require sensory input. This means that the robot may be able to have fewer sensors, which may reduce the weight, complexity, and monetary cost of the robot. Additionally, open loop control may require less processing power than feedback control since open loop controllers do not need to process sensory data.

The definition of a robot from chapter one states that a robot can sense its environment and perform action based on its sensory input. This means that a robot that only uses open loop control is not really a robot. However, a robot could use open loop control for certain tasks. Open loop control may be suitable for tasks that are very simple and predictable. One example would be a vehicle that moves along tracks to deliver cargo. It could use feedback control to load the cargo onto itself, then use open loop control for driving along the tracks for a certain amount of time before switching to feedback control again to unload some cargo. Driving along tracks would be a suitable task for open loop control since the robot would not need to navigate, it would simply need to keep driving and the tracks would take it where it needs to go. Since the open loop controller cannot use sensory data to sense when it has arrived at the correct location, the driving time between the load and unload locations would need to be very predictable. Also, it would probably be safer to use feedback control even when simply driving along the tracks so that the robot could detect any objects or people blocking the tracks.

Sensor error

Sensor error may be corrected by calibrating the sensor. If the error is consistent, it may be possible to correct the error during the processing of the sensory data. For example, if a distance sensor is consistently off by 5cm, the error could be corrected by simply adding or subtracting 5cm from the sensor value. The chances of having a sensor error might be reduced by adding more redundant sensors. For example, if a robot has three distance sensors facing the same direction, and one of the sensor values is different than the other two, it may be safe to assume that the two similar sensor values are more correct. However, adding redundant sensors can be expensive and add weight and complexity to the robot. Sensor error could also be corrected by replacing the sensor, or it could be avoided altogether by using open loop control rather than feedback control, since open loop control does not require sensory input.

If the robot can predict the sensor value, the robot could check the sensory input against its prediction and use the predicted value if the sensor value is far off. For example, a robot driving along a track to deliver cargo might know approximately how much time should pass before reaching the unload location. If the robot still has not sensed the unload location after the predicted time has already passed, it could go back to where it predicted the unload location to be and unload there.