Modeling and control of dynamic legged locomotion are more difficult than the traditional types of locomotion in that, a dynamically stable locomotor renders the stability only cyclically. This feature requires a larger prediction horizon for control and stability analysis, which complicates matters. Researchers have approached this class of problems from different points of view, which may be categorized into methods using the "true" model of the system and methods that consider a template (simplified) model. The study of legged locomotion using template models, such as LIP, SLIP, Compass Gait, and Cart-Table models, provides insight into the fundamental principles that underlie legged locomotion. The resulting platform-independent reductive models typically admit analytical representations that are more tractable than the original higher-order robot models. Significant achievements in the locomotion skills of dynamic robots have been made following both perspectives briefly mentioned above. However, there is still much to be done to meet the agility, efficiency and speed demonstrated by animals and humans, when it comes to accommodating real world environments. Are we prepared for the next Fukushima-like incidents? The workshop will discuss recent achievements in dynamic legged locomotion in robotics and biomechanics communities, with a focus on the role of simplified models.
Topics of Interest
The questions to be addressed in this workshop include, but are not limited to:
- What is the state-of-the-art in dynamic gait template models and their control?
- What is the state-of-the-art in control of legged robots based on the gait templates?
- How are issues like kinematic singularity addressed when mapping template behaviors into real robots?
- What are alternatives to template-based control of dynamic legged locomotion and how do these alternatives compare in terms of performance?
- Does the study of gait templates inform a better understanding of locomotion in animals?
- What next milestones are required for template-based approaches to advance the understanding and control of legged locomotion?
- Bioinspired dynamic locomotion gaits
- Mathematical modeling of human/animals locomotion templates (event-driven cyclic hybrid systems)
- Efficient solutions to dynamics and optimization-based control of such systems
- Comparisons of different approaches to control of dynamic gaits
- Adaptive/Reactive gait generation for dynamic bipeds/quadrupeds
- Natural dynamics exploitation in gait design and control
|Name||Affiliation||Title of talk (tentative)|
|Jonathan E Clark||FSU, US||Dynamic locomotion in horizontal and vertical domains: templates, scaling, and adaptations for multi-modal behaviors (Abstract)|
|Avik De (Daniel E Koditschek)||UPenn, US||Template Composition for Synthesis of New Behaviors from Simpler Constituents (Abstract)|
|Roy Featherstone||IIT, IT||Balancing Made Simple (Abstract)|
|Shuuji Kajita||AIST, JP||Spatially Quantized Dynamics, a new approach for model discretization (Abstract)|
|Sangbae Kim||MIT, US||Model Predictive Control for Cheetah 3 (Abstract)|
|William Martin (Hartmut Geyer)||CMU, US||Globally Optimal Gait Transition Policies for Compliant Bipedal Systems (Abstract)|
|Uluc Saranli||METU, TR||Accurate Modeling, Control and Embedding of Templates for Planar Running (Abstract)|
|Andre Seyfarth||TU Darmstadt, GE||Design and control of locomotor systems based on locomotor sub-functions and sensor-motor-maps (Abstract)|
|Tomomichi Sugihara||Osaka University, JP||To embed biped motion skill into inverted pendulum model (Abstract)|
This workshop is supported by the following IEEE Robotics and Automation’s Technical Committees:
- Human Movement Understanding
- Bio Robotics
- Wearable Robotics