Wheg Robotics
Robotic wheels have physical limitations, one of which is stairs. How often has your radio-controlled car gotten stuck when it has hit some rocks? Wheels have single-axis rotation, thus energy-efficient but restricted in terrain. On the other end of the spectrum, we have legs that come with their limitations. Aside from the complexity of control, there is more power consumption and higher development expense. Much study to find a middle ground between these two locomotion methods has been done, resulting in the wheg. It combines the energy-efficient single-axis rotation, like a wheel, but uses a claw-like design to get the climb of a leg.
Bio-inspired robotics looked into the cockroach and how it charges at obstacles and uses synergies of arm movement. The paper 'Comparing cock-roach and whegs robot body motions' explored this concept and how a wheg robot could replicate this movement.
Non-bio-inspired approaches looked more into the locomotion and engineering side, such as the European Space Agency project PROLERO. ESA designed a robot that used a similar idea to the wheg. Rather than the wheel design, it used rods on a rotating axis.
There were three designs of wheg robots we made (the first two can be seen within Figure 1). The first used an aluminium profile structure and continuous rotation servos on each wheel (seen on the left of the image).
The robot on the right of the image was an improved version where a bend in the back was implemented, alongside suspension.
Our final design built on the concepts introduced in the second robot but introduced a better design of wheg, and a stabilizer for support.
Wheg design
There are many ways to build a wheg. Its main component is a single-axis rotation and some form of the claw. PROLERO managed this with a single rod; then, some complex versions can change from a wheel to a wheg, such as the TurboQuad.
In the first design, we explored the front claw style. This design was easy to design and easy to print. The aim was to see how little we could get away with. What is the point of making a complex wheg design which does the same job as a simple one?
This wheg design did not work well as it only worked in one direction, causing the robot to get stuck in reverse. The following design used a bi-direction technique to climb out of obstacles and into them.
The designs were 3D printed, making them light and still reasonably durable (though the whegs used them within an outdoor testing environment).
Design choices
The back bending was taken from the bio-inspired nature where insects can bend their backs to increase their climb. On the wheg robot, we added a servo to allow a bend in the back. We found that this allowed the redistribution of weight which increased the climb of the robot. The first design worked well but was limited to higher climbs.
You can see how the flat back robot struggles to increase its climb on the higher rocks. The robot's upside was thin enough to drive upside down and light enough to travel over obstacles.
The following design used suspension to absorb shock. The previous design would get stuck over some obstacles and spin the wheels in the air. The springs gave the wheel more chance to get a grip.
As seen, there are limitations with the more considerable amount of weight. This meant challenges were not visible in the lighter version one model, which is more visible now.
The final model used the bi-direction wheel and used a stabiliser to increase the climb further when the back moves. This version was much better at terrain navigation.
Electronics and programming
The robots all used the Raspberry Pi Zero. This board is lower power, low cost and low weight. It ran Python as a startup script to search for a wirelessly connected Xbox controller. The code would read the command codes sent from the controller and act accordingly.
The battery was an external mobile phone charger. It provided enough current and the correct voltage for the Pi. The USB connections made it even easier to integrate with the Pi. One of these batteries was solar recharged, making the robot environmentally friendly.
The code can be found above on GitHub.
Conclusions and future work
The wheg gave an impressive performance over terrain that a wheeled robot of similar size would struggle with. An optimum wheg design is yet to be researched into but would greatly improve the development of wheg robotics.
The future work on this robot would be to make it autonomous. There are many optimal ways for the robot to bend its back to get over terrain. The robot could assess the landscape and climb over it using movements it has learned through experience.
