Skip to Content

State estimation for the path-Following control problem of visual unmanned ground systems

The state estimator for visually-guided car like mobile robots has the objective of analyse the measurements of provided by the unmaned ground platform and calculate the better estimation of the state of the vehicle to send it to the controller who closes the loop. This is one of the main systems of the Path following control system for visually guided unmanned ground systems.

It has six inputs. The measurement of the speed of the vehicle, given by the speedometer; the measurement of the angle of the steering wheel provided by an encoder; the measurements given by the visual system: the measured distance to the path, the angle between the tangent of the path and the vehicle and some punctual codes that have information about the circuit. The last input is all the available information about the circuit to follow.

In our problem, the path to follow is a painted one, but we also have to follow a magnet path with no feed-back information. The painted path is painted so that if the vehicle goes over that path, the magnet path is followed attending to the kinematics of the vehicle.

The outputs of the state estimator are many. The most important one are those used for the controller who close the control loop and are: the estimated state of the vehicle over the circuit (distante to the path, angle between the path and the vehicle and advance coordinate of the path); the estimated curvature of the path; the estimated angle of the wheel; the estimated speed of the vehicle; and the position of the vehicle in the X-Y space.

The state estimator is composed by five modules:

Fig. 1: State estimator scheme
  • Wheel position estimator
  • Speed estimator
  • Position over circuit estimator
  • Curvature estimator
  • X-Y position estimator

This state estimator uses multiples Asynchronous Extended Kalman Filters that is other CVG research line.

The performance of this state estimator was verified thanks to multiple simulations and tests in our Citroën C3 Pluriel prototype.