Contributions published at Artificial Intelligence Lab

The present thesis documents the development of a portable pneumatic system for prosthetic hands separating mass of the actuators from the point of force exertion by using tendons. Most of today's myoelectric prostheses use servo-motors to drive the hand and wrist. Our system aims to be portable, low-energy, low-noise and lightweight, but provides a high dexterity and output force while maintaining precision of movement for activities of daily living (ADLs). Pneumatic muscles are similar to biological muscles in terms of compliance while providing a high output force. As force is a crucial component to a prosthetic hand, the integration of a pneumatic system could definitely provide a new way to actuate prosthetic devices.

null

null

null

In the field of knee exoprostheses, active knee actuation is paramount to enable natural gait. This requires a suitable control strategy. A technique called "Complementary Limb Motion Estimation" uses the sound residual leg to imitate its motion on the prosthesis. The current measurement setup requires a bulky system to be installed on the sound knee, which is not suitable for clinical application. We have developed a computer vision approach using a Dynamic Vision Sensor (DVS) as a replacement. Two stages have been treated; the extraction of line features from the DVS and the development of a knee tracking algorithm.

There has been an increasing interest in applying biological principles to the design and control of robots. Unlike industrial robots that are programmed to execute a rather limited number of tasks, the new generation of bio-inspired robots is expected to display a wide range of behaviours in unpredictable environments, as well as to interact safely and smoothly with human co-workers. In this article, we put forward some of the properties that will characterize these new robots: soft materials, flexible and stretchable sensors, modular and efficient actuators, self-organization and distributed control. We introduce a number of design principles; in particular, we try to comprehend the novel design space that now includes soft materials and requires a completely different way of thinking about control. We also introduce a recent case study of developing a complex humanoid robot, discuss the lessons learned and speculate about future challenges and perspectives.

null

"Place cells" are neurons discovered in the hippocampus of rats with a high correlation between their firing rate and the spatial location of the organism in a limited area. Our work was related to this discovery but instead of a rat, an artificial neural network with a slow feature analysis learning algorithm is implemented. The goal was to create a toolbox which permits an experimenter to conduct experiments in order to build respectively to discover so called "terrain sensitive cells", i.e. cells with a high correlation between the terrain characteristics and their firing rate.

Soft robots possess several potential advantages over traditional articulated ones and have attracted significant interest in recent years. However, to control this new type of robots using conventional model-based robotic control approaches is generally ineffective. In this paper, we investigate the challenge to embed and switch among multiple behaviors for an octopus-inspired soft robotic arm. An online learning method for reservoir computing is exploited for this task. This online learning method does not require a separate teaching data collection phase; thus, it has the potential to achieve autonomy in soft robots. Our result shows the feasibility of this approach.

null

null

null

null

null

null

null

null

null