Towards dynamically running quadruped robots: performance, scaling, and comparison

The field of robotic research applying legged machines and robots is growing quickly, but it is still facing a major discrepancy of performance, quality, and complexity, compared to their biological counterparts. A lack of appropriate off-the-shelf hardware platforms, a lack of open-access to dynamically running legged robot platforms, and missing comparative studies between existing quadruped platforms (mis)leads researchers to re-invent and re-design own quadruped robots, from the scratch. As research of legged locomotion should be the goal, and not research about its tools (the robots per se), this can be a frustrating and time consuming process. Research in dynamic legged locomotion would benefit from free and easy access to blueprints, and studies where different robot design choices can be tested and compared rigorously against each other. A future, better-performing platform could be assembled from the best-performing components (actuators, sensors, driver code, controllers) and principles (type of actuation, type of control, sensing, leg design, trunk design, compliant design) of existing systems.