 Hello there, my name is Michael Fennell from the Intelligent Sensor-Actuator Systems Lab at the Karlsruhe Institute of Technology and in this video I will present our paper, Optimization with Design of a Kinesthetic-Haptic Interface with Human-like Capabilities. As you might know, a lot of kinesthetic-haptic interfaces were designed in the past. Having these in mind, we ask ourselves the questions, is there an optimal and universal design for a haptic interface? In literature, we found different approaches. For example, the maximization of a manipulability index, a task-dependent optimization framework and a detailed study of the human arm movements in activities of daily life were suggested to obtain an optimal manipulator design. However, we found that they were missing a lowest common denominator, which is the human arm itself, including its achievable poses, velocities and accelerations. To come up with the manipulated design based on this fact, we first modeled the human arm as a serial manipulator with seven joints, taking into account angle limits as well as velocity and acceleration limits. With this model, we performed a Cartesian workspace analysis where we examined the reachability on a grid of poses as we can see here. For each pose, we then examined the velocity and acceleration capabilities of the human hand by approximating the set of possible Cartesian velocities and accelerations with an ellipse as shown here. We then formulated an objective function. The goal of this function is to adjust the parameters of the manipulator so that the capabilities are matched as best as possible. This means we need to maximize the fraction of reachable human arm poses, the fraction of reachable velocities and accelerations, as well as the fraction of the desired feedback force. At the same time, our solution must satisfy the dynamic constraints of the manipulator and choose from a set of available actuators. The resulting optimization problem is a non-linear, non-convex mixed integer problem. Using a genetic algorithm, we found that there exist multiple local minima that represent different trade-offs. One of these trade-offs is shown here. We have very high position coverage, almost in the whole workspace of the human arm. The velocity coverage is also quite high, especially in the area in front of the torso. This comes at the coast of reduced worst case acceleration as we see here. To verify the results of the optimization, we created a virtual prototype. This prototype confirms that we have a very high reachability for slow movements, as we can see on the green indicators here. As expected, the acceleration indicators sometimes turn red for fast movements. This is also true for movements close to the workspace limits. Let's conclude our paper. We presented a method for analyzing the capabilities of the human arm. With this method, we created a novel optimization-driven design approach that tries to match the human arm as best as possible. Using a virtual prototype, we were able to verify the properties of the found design for real arm movements. Thank you for your attention.