ANR SHERPA Imprimer

The Sherpa Project, intelligent biped for load carrying, is designed as a first step towards a national ambition in humanoid robotics. Hence, the context of the Sherpa project must be seen from three points of view, (1) scientific, (2) strategic, and (3) economic.From the scientific point of view, this point will be addressed more deeply in the following sections, this project aims at showing the interest of a robotics device, an intelligent biped for load carrying, designed with characteristics close from the ones of living beings, and by imagining Sherpa, from start, as an artificial entity equipped with bio-inspired control methods based on a large set of sensors. We claim that, only a design radically different from the ones followed by classical “humanoids” projects (mechanical aspect inspired by classical industrial robots, “intelligent” aspect ignored until the end of the project, applications for the robot ignored) will allow to go further in terms of performances and acceptability.
Regarding the national research strategy, this project is in line with the efforts produced by the CNRS concerning the JRL9: Hence, in both places, Japan and France, the JRL aims at increasing robots’ autonomy and uses HRP-2, a 100%-japanese humanoid experimental platform, to test its algorithms. In complimentarily with the JRL, and the French humanoid community focused on the Japanese HRP-2, we wish to give the country a competence in that field. For practical considerations, we propose to focus on the design of an intelligent biped robot, instead of a full humanoid robot.
From the economical point of view, studies made in Japan by the METI10, and in the rest of the word by the United Nations11 are similar: The Service Robotics’ market will exceed the one of Industrial Robotics, will contribute to economic growth, and will become a real industry (it is the METI point of view). Up to now, this market concerns only Domestic Robotics (vacuum cleaners, mowers, etc) and we contribute to this market12. We have bigger ambitions for the future, and we think that humanoid robotics is as immature as aeronautics was during Blériot’s living. Everything is still to be done, all ambitions are pertinent.
There are many goals, but they can be synthesized in two groups: (i) a very large but coherent set of basic research activities, and (ii) a demonstrator (named Sherpa) to experiment each activity (Remark: A early prototype will be built at the mid-project to test the algorithms in real life). The basic scientific goals are the following:
1. Developing mathematical tools and mechanical design methods concerning hybrid serial-parallel mechanisms, redundant actuated mechanisms, co-actuated mechanisms, and bio-inspired mechanisms;
2. Developing robust control laws able to deal with actuation redundancy, co-actuation, and synthesized with bio-fidelity care;
3. Using omni-directional vision as the entry point of balance stabilizing algorithms, target tracking, and obstacle avoidance.
Every objective relies on several experts of the Lirmm Robotics Group, and on the experts of the Mouvement et Perception laboratory, these experts being specialized in physiological and neuronal biomechanical factors influencing motion. The experts of the Mouvement et Perception laboratory will contribute to the project from two different aspects: (i) as experts for researches made at Lirmm (concerning mechanical, control and vision activities), and (ii) by doing own pieces of work (concerning the mechanical optimizing of Sherpa from biological considerations, per example). Naturally, we will be open to other collaborations, for scientific aspects (per example: with Université Montpellier I regarding bio-mechanics), for technological aspects (per example, with the Fatronik Foundation in Spain for mechanical engineering), for economical aspects (for example, with Wany Robotics, regarding market expectations).
Finally, all these elementary research activities will be merged in building a human-scale biped prototype, able to evolve in a laboratory-type environment. For 15 years the Lirmm Robotics team has started building prototypes, particularly for European projects, and is ready to take up this challenge.

 

Participants :

 

SHERPA biped walking robot