In this report, the design of a modular test bed for elastic drive concepts which drive human mo- tions is presented. First, different elastic drive concepts are covered which are derived from me- chanical models of human muscles. Then, the human motions, in particular jumping and hopping, which are driven by these muscles, are described. Although many different muscles must be consid- ered in human motion, a simple mechanical model with only one actuator as muscle can be derived.

Considering the biomechanical knowledge, different concepts for the demonstrator are discussed. These differ in the way in which the human motion is approximated (one-dimensional and two- dimensional) and the way in which different elastic drives can be integrated. While the one- dimensional model of the jumping motion is more abstract, it can be integrated with the actuation in a serial set up. The two-dimensional motion is realized using a two-link mechanical leg running in a linear guidance. The elastic drive can be realized by either using translational or rotational de- grees of freedom; however, it is pointed out that the translational set up is easier to accomplish. The translational design is used together with a two-link mechanical leg because it better serves the demonstration purpose of the test-bed and reveals more information for possible applications in prosthetics. The leg is driven by the fixed actuator using a Bowden cable.

The power and transmission requirements for the drive chain are derived using an inverse dynam- ical model of the demonstrator which is fed by measured data of a human continuous dynamic jump. In order to meet the geometrical constraints and greatly reduce power requirements, the leg is designed using a 1:2 scale. The time evolution of the actuation power, speed and torque is de- rived. The peak power requirement for driving a mass of 2 kg is 30 W without frictional effects and the inertia of the drive train. However, these effects must still be considered for the selection of the actuator. However, this value can be compared to the literature which states a power of 15-20 W per kg of body weight, thus showing a good resemblance. Further, as the simplest elastic drive set up, a series spring is considered and optimized for the demonstrator in order to minimize the peak power and energy consumption of the actuator. The optimization towards peak power is more effec- tive, reducing the power by about 30 %.

In the design process, the drive chain is designed so that the power and transmission ratio require- ments are met. The design revealed that because of the large number of transmissions and guidanc- es, the mechanical losses are very high and the friction also affects the control of the system. This is clearly a drawback of the design, which could be overcome by using a direct drive instead of a fixed actuation box. However, the modularity is rated higher than the efficiency and can only be realized in this way. The design of the complete systems and the vendors of the mechanical components, the actuator, the sensors and the control hardware are discussed and the product costs are calculated. The appendix shows the technical drawings and the list of parts.

The next step is the acquisition of the components and the assembly of the demonstrator. As further research, some proposals can be made:

• Identification and quantification of the friction
• Design and analysis of control algorithms with friction compensation, especially for Bowden cables
• Influence of the friction and drive chain transmission on the spring stiffness for power minimization
• Determination of the parameters for various elastic drive concepts, e.g. based on the Hill and Hill-Häufle model, using optimization techniques

indexmenu_n_7