Dynamic Walking Robots and the W Prize [Industry/Research News]

R obots have been demonstrated to walk, crawl, or climb on uneven terrain. But legs and other features that might help to negotiate stairs or stepping stones come at a disadvantage: They rarely compete well with wheels for moving economically on smooth ground. While no obvious principles generally require that economy be traded for agility, the prospective robot designer faces many challenges in combining the two in a single machine. These challenges have been made explicit by the announcement of an award—the W Prize (see Figure 1)—to the first robot able to cross a specified obstacle course with both speed and efi-ciency. Prize author Tad McGeer aims to spur new innovations in robotics and control with this challenge. The challenge was borne out of the experiences of McGeer and others in designing dynamic walking machines.With these machines, walking gaits emerge automatically from the passive, pendulum-like dynamics of the legs. McGeer, inspired by chil-dren's toys of a century before, demonstrated a machine that could walk down a gentle slope with no external control or actuation (McGeer, 1990). With proper design, stability can be achieved through dynamics rather than control. These concepts carry over to more recent, actively-powered machines that walk on the level with unprecedented economy (Collins et al., 2005). Here, minimal actuation aids economy, but also leads to shortcomings in robustness. Existing machines topple easily even from the slightest of perturbations. Active control will prove necessary for robust stability. The W Prize challenge requires this robustness while retaining and even improving on economy. The eventual winner need not dynamically walk nor even have legs, but it will certainly outperform today's state of the art. The W Prize is but one of the developments guided by dynamic walking. It has become a small research community of its own, cutting across disciplines such as nonlinear dynamics , computer science, biomechanics, and even physiology. The latter interest stems from findings that humans harness passive limb dynamics (e.g. Kuo et al., 2005), sharing many of the benefits gained by the walking machines. The W Prize was announced at the First International Dynamic Walking Workshop , held in Ann Arbor, Michigan USA from 4–8 May 2006. There, more than 170 participants from five continents presented not only locomoting robots, but theoretical control algorithms, dynamical principles, and applications to human rehabilitation, including prosthetics and orthotics. They also outlined unsolved problems, many centering on designing for stability …