Refine
Year of publication
- 2025 (1)
Document Type
- Doctoral Thesis (1)
Language
- English (1)
Has Fulltext
- yes (1)
Is part of the Bibliography
- no (1)
For most humanoid robots, falls are the predominant limiting factor affecting their applications and autonomy to move in irregular environmental conditions. A similar susceptibility to falls can also be observed in humans during various daily activities due to balance control deficits or neuromusculoskeletal disorders. Despite the remarkable adaptability of the human locomotor system, as well as recent developments in robotics, balance loss and falls persist. To gain new insights into the complex interactions between bipedal locomotion and balance loss, we linked humanoid robots and humans into humanoid systems and focused on the accurate monitoring of the locomotor segment’s kinematics. This is an essential component for detecting and assessing balance disturbances, thereby improving the robustness during bipedal locomotion. Therefore, this thesis focused on the development of an inertial measurement cluster for direct kinematic measurements of, i.e. omitting numerical differentiation, a mathematical process that greatly amplifies single noise. However, despite the aim to increase the resilience of the bipedal locomotion and thereby reduce the fall risk in humanoid systems, the methods applied in this thesis had to differ in addressing the detection and assessment of balance disturbances. Therefore, the thesis comprises three sets of studies concerning the fields of humanoid robotics, humans, and sensor uncertainty assessment. In the first theoretical study, we introduced the mathematical concept of the inertial measurement cluster with special emphasis on its impact on providing sensory feedback on detected situational loss of balance during bipedal locomotion in humanoid robots. This was achieved through a kinematics-driven framework based on robust inverse dynamics evaluation and the reduction of numerical differentiation in critical terms by directly measuring the angular acceleration vector. Subsequently, we proposed a sensor fusion algorithm to estimate both the magnitude and application line of externally applied forces on robots in theory. In the second set of studies about humans, we addressed the remote detection and assessment of trip and slip events. Therefore, the sensitivity of the proposed wearable sensor-framework system (inertial measurement cluster combined with an evaluation framework) to automatically detect balance disturbances was examined. We were able to automatically assess the balance recovery performance of individuals and resolved the well-known adaptation phenomena to repeated trip-like perturbations. Subsequently, we expanded the functional scope of the wearable sensor-framework system and provided evidence of its high accuracy in detecting and classifying balance disturbances during simulated activities of daily life. In the third set of studies, we established a multi-method framework to provide an experimental angular acceleration reference to objectively quantify the measurement uncertainty of the proposed inertial measurement cluster. Moreover, we confirmed the reference can serve as a measurement standard. Finally, based on the measurement standard, we proposed a concept for an adjustment routine to compensate for the deterministic errors of the inertial measurement cluster and confirmed a measurement uncertainty reduction. In conclusion, we established a sensor suitable for humans and humanoid robots, omitting numerical differentiation, and confirmed its significantly reduced measurement uncertainty. The proposed framework approaches based on the inertial measurement cluster were the key factors for the accurate detection of balance disturbances in humanoid robots as well as humans, highlighted by comparisons to conventional methods based on numerical differentiation. Consequently, the proposed sensor and frameworks have the potential to provide new insights into the causes of balance disturbances or factors that lead to insufficient reactive actions to prevent falls in humanoid robots and humans.