Embodied Interaction as a Core of Cognitive Interaction

Embodied Interaction as a Core of Cognitive Interaction: A holistic approach towards autonomous walking systems
Project Acronym: 
EICCI
Duration: 
01.10.2013 until 31.10.2017
Summary: 

Animals as small as insects easily outperform today’s technical devices in terms of (1) computational and energy resource efficiency, (2) adaptive dexterity across contexts and tasks, and (3) compliant, durable, and well-controllable mechanics. The project EICCI aligns the expertise of two engineering and three neuroscience groups to develop a demonstrator that, with regard to the above criteria, tackles today’s technical limits and integrates state-of-the-art research on biomechanics, sensory information processing, and context-/task-dependent motor-control. 

EICCI is a CITEC-only sequel based on the CITEC-coordinated EU-FP7 project EMICAB (www.emicab.eu). It is centered around the biomimetic research platform HECTOR, a compliant, highly sensorised, six-legged walking robot. HECTOR is one of CITEC’s few own developments in robotic demonstrators, and an ideal integration platform for research on different topics in computational neuroethology and biomimetics. The overall goal of EICCI is to demonstrate bio-analogue autonomous behaviour in real time and in outdoor environments, testing hypotheses on embodied interaction based on neural motor control and multisensory perception. Specifically, we are integrating research on control of multiple compliant actuators, adaptive coordination of limbs, multisensory exploration, and internal planning of complex motor actions.

Scientific Goals: 

EICCI addresses all three research themes of the CITEC research agenda on Motion Intelligence (Area A):

  • Resource efficiency: Developing an efficient, bio-analogue motion vision system and validating the robustness of motion-based navigation algorithms. 
  • Physical interaction and proactive behaviour: Revealing mechanisms of distributed load processing in adaptive coordination of walking. 
  • Skill acquisition: Developing a framework for the dynamic internal representation of movement. 
Work Packages: 

WP1 Vision: This WP targets the integration of an active vision module into the HECTOR environment. It combines the realization of a novel hardware platform for a panoramic insect-inspired visual system with the development of models for processing the visual image flow and for mediating visually guided orientation behaviour.

WP2 Proprioception and compliance: The aim of this WP is to transfer conceptual understanding of distributed load/strain sensing from a walking/interacting insect to a walking/interacting robot. 

WP3 Simulation: This WP is concerned with the development of simulations to optimally model results from biological experiments and to model mechanical, control, and networking aspects of HECTOR. 

WP4 Mechatronics and integration: In this WP, the various mechatronics subsystems of HECTOR, including strain sensors, the vision module, and drives connecting the main body segments, will be tested and integrated.  

WP5 Benchmarking and evaluation: This WP evaluates the newly developed concepts in a real-world environment. It evaluates the resource efficiency of the bio-inspired vision module, the significance of distributed load sensing for adaptive control of walking, vision-guided walking, and high-level motion control from multisensory information.   

Selected publications of the project: