Putting Their Best Feet Forward

Legged robots have long been a source of fascination for scientists and engineers, offering the promise of versatile locomotion across a wide range of terrain types. However, designing such robots has proven to be a formidable challenge, as they must be equipped with advanced sensors and software to navigate these complex environments.

One of the primary challenges of designing legged robots lies in ensuring that they can traverse a wide range of terrains, from flat surfaces to rocky terrain, sandy beaches, and even steep inclines. This requires the robot to be able to adjust its gait, adapt to changes in surface conditions, and maintain its balance at all times.

To enable effective locomotion, robots rely on a range of sensors, including cameras, LIDAR, and inertial measurement units (IMUs), to collect data on their environment and adjust their movements accordingly. Cameras are used to provide a visual representation of the environment, while LIDAR systems can provide precise distance measurements and detect obstacles in the robot's path. IMUs, which measure changes in the robot's acceleration and orientation, are used to maintain balance and adjust the robot's gait.

However, these sensors add a significant level of complexity and expense to the design and operation of legged robots. In addition to the sensors themselves, sophisticated algorithms and software are needed to process the sensor data and make real-time decisions about the robot's movements. These factors impose heavy limitations on who can deploy such robots, and for what applications.

After observing the radically different means of locomotion of centipedes, a team of researchers at the Georgia Institute of Technology wondered if robots might also benefit from employing similar mechanics. Centipedes, with tens to hundreds of legs, are well known for their ability to traverse just about any type of terrain without even slowing down, after all. The team’s belief was that by having redundant legs, a robot might be able to easily traverse difficult terrain without requiring any special sensing or motion planning algorithms.

To test out their idea, several robots were constructed with varying numbers of legs. Starting with six, then increasing by two at a time, all the way up to sixteen legs, these robots were observed as they attempted to navigate rough, bumpy, debris-ridden, or otherwise complex terrain. The robots had no special sensing capabilities or advanced navigation algorithms, and instead relied on what the team referred to as the spatial redundancy of their legs.

This spatial redundancy allowed the robots to continue moving along their course even if one, or a few of the legs faltered, because the others still had a firm foothold. The simplicity of design also makes the robots more reliable by eliminating sensors that can be finicky or fragile. This reliability could be especially useful in harsh environments, such as those that will be encountered during planetary explorations.

At this point, the researchers have only shown that increasing the number of legs helps robots traverse difficult terrain, but they have not yet proven what the optimal number of legs is. They are working to understand this at present, and this work should help in understanding the tradeoffs between energy consumption, speed, power, and robustness for future many-legged robots.

The researchers have already put their findings to work in a startup business in the agriculture industry. They are using their robots to remove weeds in fields where traditional herbicidal agents are ineffective.