Stefano Miani - Ongoing Activity

Robot motion control in a constrained environment (F. Blanchini, S. Miani, F.A. Pellegrino)

The robot end effector has to move in the shown environment without hitting any of the objects.
Given the specifications of the objects, a proper representation of the workspace is first computed off line (see Figure 1) along with a set of via points (the green dots in Figure 2) and a minimum distance tree (this part of the work was developed by Bart Van Arkel, visiting student from the Delft Technical University, The Netherlands).

2 and 3-D view of the robot working space
Figure 1: Workspace 2 and 3-D representation

Via points, 2 and 3-D current position online interface
Figure 2: Via points, 2 and 3-D current position online interface


Then, given the current and target states, each motion is carried out on line by a two level controller according to the following scheme:

  1. the high-level controller computes the optimal sequence of via points via the minimum distance tree and passes the "next" optimal via point to the low level controller
  2. the low level controller avoids obstacle hitting during the motion.

A movie with the experimental results of such control scheme for different target values can be downloaded here (robot1.mpg, 3Mb)

Bi and tridimensional views of sample motions for different implemented low level control laws are depicted in Figs 3--8

Linear saturated control law u=sat(Kx)
Figure 3: 2D sample motion with a linear saturated control law u=sat(Kx)

Linear saturated control law u=sat(Kx)
Figure 4: 2D sample motion with a linear saturated control law u=sat(Kx)


Figure 5: 2D sample motion with a nonlinear set based control law


Figure 6: 2 and 3D sample motion with a nonlinear set based control law


Figure 7: Sample motion as in Fig 6, different view


Figure 8: Sample motion as in Figs 6-7, different view