Kawasaki has recently revealed its computer-generated concept for the Corleo, a "robotic horse". The video shows the automated equine galloping through valleys, crossing rivers, climbing mountains and jumping over crevasses.
Author
- Matías Mattamala
Postdoctoral Researcher, Oxford Robotics Institute, University of Oxford
The Corleo promises a high-end robotic solution to provide a revolutionary mobility experience. Kawasaki's current motorbikes are constrained to roads, paths and trails, but a machine with legs has no boundaries - it can reach places no other vehicles can go.
But in the case of the Corleo, how feasible is it to achieve such a level of agility and balance, while safely carrying a human through natural environments? Let's discuss what would be needed to achieve this.
A robot is a complex machine with two main components: a body and an information processing unit. The body has a particular morphology that determines the robot's function, and carries actuators (devices that convert energy into physical motion) and sensors to act in the world and understand it, respectively.
The information processing unit is usually a computer, which implements algorithms to process data from the sensors, build representations of the world and determine the actions to be executed, subject to a specific task of interest.
Simple robots, such as robotic vacuum cleaners satisfy these requirements. They have a suitable body for going under furniture and not getting stuck (their flat top is also useful to give your cats a ride).
The actuators are the motors that spin the wheels and the vacuum system. It has impact sensors to detect collisions, and some even have cameras for understanding the environment. Owners can set a cleaning routine, and the vacuum's computer will determine the best way to execute it.
The Corleo is a quadruped robot, one of the most stable legged robot configurations. The four legs seem strong and capable of flexing forward and backward to run and jump.
But they seem limited in movements known as abduction and adduction . If I push you on your right side, you will open your left leg - this is the abduction motion helping you keep balance.
Adduction is the opposite motion - a movement towards the midline of the body. Perhaps this is just a limitation of the concept design but, either way, the Corleo needs this articulation to ensure a safe and smooth ride.
Next comes actuators. Legged robots, in comparison to wheeled vehicles, need to continuously balance and support their own weight. They also provide a level of suspension that provides cushioning for the rider.
They need to be strong enough to push the robot's body forward. On top of that, the Corleo will also carry a person. While this is currently possible, such as with the the Barry robot or Unitree wheeled robots , the Corleo also aims to gallop and jump over gaps. This requires even more dynamic and stronger actuators than the previous examples.
A manually driven car or motorcycle doesn't need sensors or a processing unit, because the driver steers the car depending on what they see. But a robotic horse does need more sophisticated control systems to determine how to move the legs, otherwise we would need both hands and even our feet to drive it.
Locomotion control has been an active area of legged robotics research since the 1940s. Researchers have shown that a legged machine can walk down a slope without motors or sensors (which is called "passive" locomotion).
If only "proprioceptive" sensors - the types of sensors that tell your phone when to rotate the screen - are used to control balance, it's called "blind" locomotion because it doesn't rely on information from the external environment. When a robot also uses "exteroceptive" sensors to determine how to walk, which refers to sensors that pick up information about the environment, it's called "perceptive" locomotion . This is what Corleo shows.
From the pictures released, I could not spot any visible cameras or Lidars - laser range finders. They could be hidden, but it would be reassuring to know that the Corleo has a way to "see" what's in front of it while walking.
While it will be manually steered (so that it doesn't need to navigate autonomously), its locomotion system needs sensor data to determine how to step on rocks, or detect if the terrain is slippery. Its sensors should also be reliable under different environmental conditions. This is already a huge challenge for autonomous cars.
Challenges ahead
The Corleo is a concept, it does not exist - yet. As a product, it promises to be a more capable version of a quad bike. This can open new opportunities for transportation in remote areas, tourism businesses, new hobbies (for those who can afford it), and even sports.
But I'm more excited about the technological advances that the achievement of such a platform implies. Legged robots do not necessarily need to look like quadrupeds or humanoids.
Self balancing exoskeletons, such as Wandercraft's Personal exoskeleton or Human in Motion Robotics' XoMotion , are legged robots that are revolutionising the lives of people with mobility impairments. The technological advances implied by the Corleo could have be of major benefit to the development of assistive devices for disabled users, enabling them to achieve independence.
Current progress in legged robotics suggests that many features proposed by Kawasaki are feasible. But others pose challenges: Corleo will need the endurance to walk in the wild, run effective locomotion algorithms and also implement the safety standards required for a vehicle.
These are all major hurdles for a reasonably sized robot. If you ask me today, I'd be unsure if this can be achieved as a whole. I hope they prove me wrong.
Matias Mattamala is currently funded by an EPSRC C2C Grant at the University of Oxford in collaboration with ETH Zurich. He does not work for, consult, own shares in, or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond his academic appointment.
