Summary: Robot Locomotion

What is Locomotion

Locomotion represents the mobility mechanism of a robot in its environment. It can take many forms, such as walking, jumping, running and flying. Locomotion mechanisms are often inspired by nature, nevertheless, the wheel was a uniquely human innovation.

Mobile robots typically employ wheeled or legged locomotion. Wheeled locomotion is more common for surfaces that are flat and hard, while legged locomotion is better suited for softer or more uneven terrains.

One common issue that affects all type of locomotion is stability, which itself is dependent on other factors such as the characteristics of contact with the ground (shape of the contact point, angle of contact, friction, etc.) and the type of environment (soft/hard ground, water, etc.).

❗Holonomic vs. non-holonomic

Holonomic robots are characterized by their ability to move instantly in any direction without considering mass or inertia, thus eliminating complex mechanical controls and challenges associated with motion by, for example, easing path planning and localization.

Moreover, holonomy refers to a restriction (or not) among translational axes. Specifically, if a robot is holonomic with respect to N dimensions, it’s capable of moving in any direction in any of those N physical dimensions available to it. If it’s nonholonomic, it’s restricted in which directions it can move in.

For example, in a one-dimensional space, there is only one axis in which something can move. If you pretend that axis is a railroad track, a train would be considered holonomic because it could potentially move in either direction. If you took some object that could only move in one direction on the tracks, that object would be nonholonomic.

If, however, you expanded that space into two dimensions such that there were two axis with one still being railroad tracks, a train would no longer be considered holonomic, because it could still only move along one predefined axis, and is not capable of moving along the other axis.

From the point of view of robotics, you can refer to humans as holonomic within our two-dimensional space (we can’t fly). If you built a robot that could move in any direction like a human can, it would also be holonomic. If you built a robot that could only move forward, or sideways, or backwards, that robot would be nonholonomic.

Holonomic vs non-holonomic robots

Legged locomotion

Legged robots are able to navigate uneven surfaces, bypass holes and even manipulate objects. The main challenges related to this type of locomotion are about the power requirements and the mechanical complexity of the designs of such robots. In fact, while on one hand it is very easy for biological life to have joints that allow for legged walking, creating a mechanical structure that can do the same is a very difficult task. For this reason, most robots rely on wheeled navigation.

Another challenge in legged locomotion is managing balance, addressed through two main strategies: static and dynamic balance.

Static balance ensures that the robot’s center of mass remains within the support polygon formed by its contact points with the ground. Starting from a three-legged configuration, robots robots are able to achieve static stability, ensuring balance without movement. To allow for static walking at least four legs must be present, but for an even simpler balance control, six legs are necessary.

Dynamic balance focuses on maintaining balance through motion and it involves complex controls to prevent slipping. One way to model this type of balanced is through Zero Moment Points (ZMP), which anticipates the balancing of horizontal forces and friction to prevent falls.

Moreover, robots typically require at least two degrees of freedom per leg to achieve basic movement. Adding additional degrees of freedom improves a robot’s adaptability but it comes at the cost of increased energy demand, complexity in control and added weight.

❗Gaits

Gaits are precomputed sets of movements that simplify moving an agent by specifying the specific pattern of velocity and movements based on the terrain (which could depend on the movement of a set of legs as well as individual ones).

Examples of such gaits are:

  1. Trot, where diagonal pairs of legs move alternately;
  2. Pace, with lateral pairs moving alternatively;
  3. Bound, where front and back pairs of legs take turns moving.

The movement of jointed legs can be guided by a small number of oscillatory patterns. One technique that is used to express those patterns is the Central Pattern Generator (CPG), which are neural structures responsible for rhythmic activities such as breathing or walking, which can operate without sensory input but be adjusted in response to it. Another technique is motion capture, which replicates the movements observed in living being directly onto robotic systems.

The cost of transportation reflects the energy used per distance, as well as some dissipation, which takes the form of friction through rolling resistance or foot-ground interaction. Legged systems consume more energy due to their accelerations and deceleration phases.

To relieve some of the challenges involved in these kinds of inefficiencies, a solution is to leverage the dynamics of the mechanical structure of the robot itself to facilitate efficient movement.

To showcase this kind of dynamics, “passive walkers” robots have been invented, which can move down inclines with minimal energy loss. In practical terms, the robot uses forward falling combined with passive leg swing to swiftly adapt to changes in the terrain.

However, the benefits of exploiting mechanical dynamics are velocity-dependent and cannot be used in every circumstance.

Examples of legged robots

One-legged robots can navigate through rough terrains more efficiently than their multi-legged counterparts. For example, the Raibert hopper uses hydraulic actuation for movement but it requires an external power source.

Raibert hopper robot from MIT
Raibert hopper robot from MIT

Two-legged robots unlock the capabilities of running, jumping and climbing stairs by using high power-to-weight ratios.

Four-legged robots potentially offer the most amount of benefits in human-robot interaction, especially for their ability to carry heavy loads over rough terrains.

Six-legged robots have yet to match the efficiency and versatility of their biological counterparts, but having many legs leads to an increase in the gripping strength.

Wheeled mobile robots

Wheeled robots are the simplest and most efficient mechanical structure that doesn’t face major balance issues. Just like legs, a minimum of three wheels is required for stable balance, though two-wheeled robots can also achieve stability.

3 wheel arrangements

❗Differential drive

2 wheels independent of each other, where the relative velocity between one wheel and the other translates to steering.

When using more than three wheels, a suspension system becomes necessary to navigate uneven terrains.

We are going to focus mainly on the challenges related to traction and stability, as well as maneuverability and control.

Firstly, we are going to categorize wheels into four major classes:

  1. Standard wheels: Traditional circular components designed for straightforward linear movement and stability in various applications.
  2. Caster wheels: Swivel-mounted wheels equipped with a rotating mechanism, offering enhanced maneuverability and directional control.
  3. Swedish wheels (omnidirectional wheels): Specialized wheels featuring a unique design aimed at providing optimal traction and stability, often utilized in healthcare equipment.
  4. Spherical wheels: Innovative ball-shaped wheels capable of moving in any direction, facilitating smooth and versatile motion across diverse surfaces.
different wheel types
a: standard wheels, b: castor wheels, c: Swedish wheels, d: spherical wheels

The choice of wheels to be used in a robot is linked to maneuverability, controllability and stability.

Stability requires a minimum of two wheels, but is guaranteed by having three and improved by having more. Maneuverability is enhanced by using omnidirectional wheels and controllability tends to inversely correlate with maneuverability.

Different types of steering and wheel configurations:

We can distinguish between four different types of wheel arrangements, which lead to different types of steering:

  • Synchro drive: three driven and steered wheels controlled by two motors;
Wheel configuration for synchro drive
  • Omnidirectional drive: allow for movement in any direction;
Configuration of wheels for omnidirectional drive
  • Tracked slip/skid locomotion: similar to a tank, it provides excellent maneuverability and traction in loose or rough terrain;
tracked-locomotion
Nanokhod Robot
  • Akerman steering: this kind of steering ensures that the inner and outer wheels of a turning vehicle follow distinct arcs. Specifically, the inner wheel pivots more sharply than the outer one, preventing tire scrubbing and stress on the vehicle’s components

💡Skidding vs. slipping

Skidding refers to the vehicle’s tires losing traction with the road surface while still rotating (eg. breaking). Slipping involves the tires spinning without gaining traction (eg. accelerating on slippery surface).

  • Walking wheels: a hybrid of legged and wheeled locomotion, allows for navigation through rough terrains.
walking-wheels
Example of hybrid robot with both legs and wheels

Other types of locomotion

Other types of locomotion replicate movements found in nature, such as vibration or oscillation. We’ll analyze crawling, sliding, running and jumping robots:

  • Crawling: inspired by caterpillars, they utilize longitudinal vibrators to move. This kind of movement is very energy-intensive;
Active Scope Camera robot
  • Sliding: inspired by snakes, it uses omnidirectional wheels for movement. The difference with crawling is in the way locomotion is carried out;
  • Running: it involves oscillatory leg movements leading to horizontal motion;
  • Jumping: similarly to running, it uses oscillatori leg movements that lead to vertical motion.

❓Check your understanding

  • Describe the difference between Ackerman and differential steering.
  • Describe the difference between holonomic and nonholonomic vehicles.
  • Rank the following list of locomotion types in order of power demands: crawling/sliding, running, tires on soft ground, walking, and railway wheels.
  • Describe the difference between static and dynamic balance and why static stability is often considered desirable.
  • Define support polygon, zero moment point, and action selection.
  • Define the role of reference trajectory and central pattern generators in locomotion.
  • List three virtual gaits.
  • What are the main differences between legged and wheeled locomotion in robots?
  • What are some of the advantages and challenges of legged locomotion?
  • What are the four major classes of wheels used in wheeled robots, and what are their characteristics?
  • Can you describe the different types of wheel arrangements used in wheeled robots and their applications?
  • How does the choice of wheels in a robot relate to its maneuverability, controllability, and stability?
Did you find this article interesting?
YesNo
Scroll to Top