Whitebox

Title: Erklärbare Modelle für menschliche und künstliche Intelligenz (Explainable models for human and artificial intelligence)

Funding:  LOEWE-Schwerpunktantrag, granted  by the Hessisches Ministerium für Wissenschaft und Kunst from 1. 2021 to 12.2024

Consortium: Nine academic partners from different departments of TU Darmstadt are involved in this project. You can see details here

I am leading Leading WP3.1 „Man and Machine walk together" in this project. Here we develop the bioinspired models of human gait and connect them to robot models.

EPA 2

Title: Integrating Locomotor Subfunctions with Electric-Pneumatic Actuation

Funding: DFG granted project from 8.2021 to 7.2024

Summary of the project:

Compared to biological muscles, current technical actuators are limited in their performance and versatility to realize human-like locomotion. For resolving this problem we need to better understand biological legged locomotion which can be described in a three-level structure: 1) generation of the different locomotor subfunctions (LSF), namely stance, swing and balance, 2) composition of LSFs for versatile legged locomotion and 3) LSF adaptation for various locomotion tasks and conditions.

In order to overcome the actuator limitations for locomotion, we recently introduced the hybrid EPA actuator as a combination of electric and pneumatic actuators. The EPA design provides direct access to the control and morphological properties. We recently demonstrated that with the EPA, the actuator limitations could be clearly reduced for stance LSF in vertical hopping.

In this follow-up project, we will explore the full potential of the EPA approach by extending its application to versatile locomotion following the above-mentioned three levels. First, we want to understand how the EPA design and the corresponding control needs to be adapted to match different (isolated) LSFs. In the next level, we extend the EPA approach to multiple LSFs. Here we expect that the different LSFs interact in a modular way with a parsimonious exchange of sensory information. Finally, we will study the required adaptation of identified EPA modules to realize different locomotion tasks and conditions.

Bioinspired Vision:

The benefits of EPA-based design and control will be validated with new bioinspired legged robots (EPA-Jumper and EPA-Walker), both modular and extendable to different body architectures and movement goals. By exploiting control embodiment (e.g., by implementing biarticular actuators), we will take advantage of the mechanical and functional properties of the human body, which can barely be replaced by using neural control.

The EPA design will be optimized to minimize energy consumption and maximize robustness against perturbations over a defined range of movement conditions. Experimental data on human walking and hopping (with optional perturbation) will be used to optimize the EPA design and control.

With the envisioned coevolution of mechanics and control design, EPA technology enables new versatile, efficient, and robust locomotor systems for a wide range of applications. For this, we provide the required infrastructure to easily switch between different gait conditions with high energy efficiency and minimum control effort.

Collaborators: 

In this project, a collaboration between Lauflabor locomotion lab at TU Darmstadt and HOSODA Lab in the University of OSAKAPoWeR Lab from Georgia Tech and Biorobotic lab from TU Delft Biorobotic lab from TU Delft is foreseen and there will be ample opportunities for collaboration with these research groups.

EPA

Title: Hybrid Electric-Pneumatic Actuator (EPA) for legged locomotion

Funding: DFG granted project from 8.2018 to 4.2022

Summary of the project:

A better understanding of how actuator design supports locomotor function may help design and develop novel and more functional powered assistive or robotic legged systems. Legged locomotion can be described as a composition of locomotor sub-functions, namely axial leg function, leg swinging and balancing.

In this project, we focus on the axial leg function (e.g., spring-like hopping) based on a novel concept of a hybrid electric-pneumatic actuator (EPA). This principal locomotor sub-function determines the movement of the body center of mass. We will design and manufacture EPA prototypes as enhanced variable impedance actuators (VIA). In contrast to other VIAs, the EPA provides not only adaptable compliance (e.g. an adjustable spring) but with the pneumatic artificial muscle (PAM) also an additional powerful actuator with muscle-like properties, which can be arranged in different configurations (e.g., in series or parallel) to the electric motor (EM). This novel hybrid actuator shares the advantages of EM and PAM combining precise control with compliant energy storage required for efficient, robust, and versatile human-like leg motions via simple control laws.

Bioinspired Vision:

Based on human experiments, the EPA design will be optimized to minimize energy consumption and maximize robustness against perturbations within a desired operational range. We consider human hopping in place as a simple movement concentrating on the axial leg function.A simulation model of human muscle-skeletal function reproducing human hopping experiment results will be used to identify the objective function for the biological actuators (muscles) through “inverse optimal control”. This biologically inspired cost function will then help us to identify the most appropriate EPA actuator design. A robotic setup of the MARCO-2 hopping robot will be equipped with EPA to demonstrate and evaluate the actuator design and control. Based on its mechanical properties and its flexible arrangement in multi-segment-systems, the EPA provides a novel actuator that mimics human muscle function and is able to mechanically adapt to different gaits and conditions (e.g. locomotion speed). Preliminary experimental and simulation studies in our group show evidence of the expected advantages of adding PAM to EM. We expect that only a limited exchange of sensory information between the different locomotor sub-function controllers will be required enabling the envisioned modular architecture of the locomotor control system. With EPA technology, new versatile, efficient and robust locomotor systems for a wide range of applications can be designed.

Collaborators: 

In this project, we established collaboration between Lauflabor locomotion lab at TU Darmstadt and HOSODA Lab in the University of OSAKA. Prof. Hosoda and three of his students visited Lauflabor o yearly bases. This collaboration resulted in building EPA-Hopper robot and several experiments. The outcomes are published in different articles. Please see  here.

BioBiped 3

BioBiped

Title: Elastische, bionisch inspirierte, zweibeinige Roboter (Elastic bioinspired bipedal robot)

Funding: DFG granted project from 2009 to 2015.

Summary of the project:

The goal of this project was to investigate a novel, biologically inspired, three-segment and elastic humanoid robotic leg concept in simulation and experiment with a view to the long-term realization of walking and running in bipedal robots with human-like efficiency. This project was a practical step to extending template models to anchor level and apply to build a bioinspired robot. The aim was to rely as much as possible on the passive stability of the walking patterns, but at the same time to enable adaptation of the motion sequence with regard to superimposed motion goals.

 

In 2010, BioBiped 1 was presented as the first of a planned series of musculoskeletal robotics platforms being developed for the purpose of investigating and evaluating hypotheses and results from biomechanics of human locomotion in robotics and their transfer to new robotic platforms.

Based on scientific and experimental results, the mechanics and electronics were modified towards a more innovative design with enhanced locomotion capabilities. Two more robots, BioBiped2 and BioBiped3, were presented in 2012 and 2014/15, respectively.

 

The BioBiped 3 robot has six actuators per leg including three  Using serial elastic actuation mimicking the main muscle groups in human legs, we aim at reproducing dynamical human gaits, namely running (jogging) and walking as well as stable standing. While the first iterations are restricted in the motion dimensions (hopping, motion in a plane), we ultimately aim at free, autonomous running robots that can perform maneuvers. For this, an onboard PC is implemented. For further information, see the project homepage www.biobiped.de.

Bioinspired Vision:

The vision of humanoid robots which mimic the abilities of humans has inspired researchers for decades. Yet transferring human abilities into a robotic counterpart has proven to be highly challenging in most cases. The BioBiped project aimed at realizing human-like locomotion. 

Integration of biomechanics research in the concept of the development of versatile, robust and energy-efficient bipedal robots may represent an essential tool to get a step closer to robots with human-like locomotion capabilities.

Collaborators: 

In this project, the SIM Group of TU Darmstadt and the Lauflabor Locomotion Laboratory of TU Darmstadt cooperated. The outcomes are published in different articles. Please see  here.