Considering Model Uncertainties for Dexterous Manipulation Using Multi-Fingered Grippers
In 2023, ESME, through a postdoctoral research project, adapted the synthesis of a controller based on the scenario approach to the case of planar dexterous manipulation using a two-finger, three-degree-of-freedom gripper. To do this, we based our work on [Esfahani 14] and applied its results to the context of dexterous manipulation.
We tested the new controller on the system shown in Figure 1.a), a 2D system with two fingers, incorporating uncertainties both in the contact points and in the initial conditions. Figure b shows results for initial condition IC1, with the reference trajectory in blue, the controller obtained through the scenario approach in red, and the controller obtained via the previous discretization-based method (used as a reference for comparison) in green. These results show that the controller derived from the scenario approach achieves better tracking performance, particularly along the y-axis. Figure c, obtained for initial conditions IC2, provides similar results. Finally, Figure d shows that for IC2, when the reference controller is synthesized via a second method (feasibility), the control diverges, and dexterous manipulation is no longer stable.
Evaluation of the scenario approach for dexterous manipulation: 
These results lead to two main conclusions:
- It is possible to apply the scenario approach to systems for dexterous manipulation. This method probabilistically guarantees system stability.
- This approach yields better results than previous methods, both in terms of stability and trajectory tracking.
In 2024, we disseminated these results through a publication.