This post will show you how to choose a robot chassis kit with wheels and motors, a motor controller, and some power for the robot, talking through the trade-offs and things to avoid.
Choosing a robot chassis kit
The chassis, like the controller, is a fundamental decision when making a robot. Although these can be self-made using 3D printing or toy hacking, the most simple place to start is with a robot chassis kit. These kits contain sets of parts to start off your robot build. A chassis can be changed, but it would mean rebuilding the robot.
The internet has plenty of robot chassis kits around. Too many, so how do you choose one?
Size
Getting the size for a robot right matters too. Take a look at the following photos:
Chassis 1 is 11 cm in and just about fits a controller in it, but is too tiny. This will make it hard to build your robot. Squeezing the controller, power, and all the sensors into this small space would need skill and experience beyond the scope of a first robot build.
Chassis 2 is Armbot. This large robot is 33 cm by 30 cm, with an arm reach of another 300 mm. It needs eight AA batteries, big motors, and a big controller. These add to the expense and may cause issues around power handling for a new builder. It has lots of space, but issues around weight and rigidity. Armbot is one of my most expensive robots, excluding the cost of the arm!
Chassis 3 in the preceding image will fit the Pi, batteries, and sensor, but without being large and bulky. It is around the right dimensions, being between 15-20 cm long and 10-15 cm wide. Those that have split levels might be great for this, but only one or two levels, as three or four will make a robot top heavy and may cause it to topple. This has enough space and is relatively easy to build.
Wheel count
Some robot chassis kits have elaborate movement methods, legs, tank tracks, and tri-star wheels, to name a few. While these are fun and I encourage experimenting with them, this is not the place to start at. So, I recommend a thoroughly sensible, if basic, wheels on motors version.
There are kits with four-wheel drive and six-wheel drive. These can be quite powerful and will require larger motor controllers. They may also chew through batteries, and you are increasing the likelihood of overloading something. This also makes for trickier wiring, as seen in the following:
Two-wheel drive is the simplest to wire in. It usually requires a third wheel for balance. This can be a castor wheel, roller ball, or just a Teflon sled for tiny robots. Two wheels are also the easiest to steer, avoiding some friction issues seen with robots using four or more wheels.
Two wheels won’t have the pulling power of four or six-wheel drive, but they are simple and will work. They are also less expensive:
Wheels and motors
A kit for a beginner should come with the wheels and the motors. The wheels should have simple non-pneumatic rubber tires. The most obvious style for inexpensive robots is shown in the following photo. There are many kits with these in them:
The kit should also come with two motors, one for each wheel, and include the screws or parts to mount them onto the chassis. I recommend DC Gear motors, as the gearing will keep the speed usable while increasing the mechanical pushing power the robot has.
Importantly, the motors should have the wires connected, like the first motor in the following photo:
It is tricky to solder or attach these wires to the small tags on motors, and poorly attached ones do have a frustrating habit of coming off. The kits you will want to start with have these wires attached, as can be seen in the following:
Another point to note is that where the motors are mounted, the kits should have some encoder wheels, and a slot to read them through. The encoder wheels are also known as odometry, tacho, or tachometer wheels.
Simplicity
You don’t want to use a complex or hard-to-assemble kit for your first robot build. I’ve repeated this throughout with two-wheel drive, two motors with the wires soldered on and steering clear of large robots, or unusual and interesting locomotion systems, not because they are flawed, but because it’s better to start simple. There is a limit to this, a robot kit that is a fully built and enclosed robot leaves little room for learning or experimentation and would actually require toy hacking skills to customize.
Cost
Related to simplicity is cost. Robot chassis kits can be brought from around $15, up to thousands of dollars. Larger and more complex robots tend to be far more costly. Here, I am aiming to keep to the less costly options or at least show where they are possible.
Conclusion
So, now you can choose a chassis kit, with two wheels and a castor, two motors with wires soldered on them, slots, and encoder wheels. These are not expensive, and widely available on popular internet shopping sites as “Smart Car Chassis,” with terms like “2WD”:
The kit I’m working with looks like the preceding photo when assembled without the Raspberry Pi.
Choosing a motor controller
The next important part you’ll need is a motor controller. Much like the motors, there are a number of trade-offs and considerations before buying one.
Integration level
Motor controllers can be as simple as motor power control driven from GPIO pins directly, such as the L298. This is the cheapest solution: a generic L298N motor controller can be connected to some of the IO pins on the Raspberry Pi. These are reasonably robust and have been easily available for a long time. They are flexible, but using parts like this will take up more space and need to be wired point to point, adding complexity to the build:
Others are as complex as whole IO controller boards, many of which hide their own controller similar to an Arduino, along with motor control chips. Although the cheapest and most flexible ways are the most basic controllers, those with higher integration will reduce size, keep the pin usage count low (handy when you are connecting a lot to the robot), and may simplify your robot build.
They often come integrated with a power supply too.
Motor controllers can be bought as fully integrated Raspberry Pi hats, boards designed to fit exactly on top of a Raspberry Pi. These tend to have a high level of integration, as discussed before, but may come at the cost of flexibility, especially if you plan to use other accessories.
Pin usage
When buying a motor controller in Raspberry Pi hat form, pin usage is important. If we intend to use microphones (PCM/I2S), servo motors, and I2c and SPI devices with this robot, having boards that make use of these pins is less than ideal.
To get an idea of how pins in different boards interact on the Raspberry Pi, take a look at https://pinout.xyz , which lets you select Raspberry Pi boards and see the pin configuration for them.
Controllers that use the I2C or serial bus are great because they make efficient use of pins and that bus can be shared.
At the time of writing, PiConZero, the Stepper Motor Hat, and ZeroBorg all use I2C pins. The Full Function Stepper Motor Hat is able to control DC motors and servo motors, is cheap, and is widely available. It also has the pins available straight through on the top and an I2C connector on the side. It’s designed to work with other hats and allow more expansion.
Size
The choice of this depends on the chassis, specifically the size of the motors you have. In simple terms, the larger your chassis, the larger a controller you will need. The power handling capacity of a motor controller is specified in amps. For a robot like the The Robot Kit I’m Using image, around 1 to 1.5 amps per channel is good. The consequence of too low a rating can be disaster, resulting in a robot that barely moves, while the components cook themselves or violently go bang. Too large a controller has consequences for space, weight, and cost:
The level of integration can also contribute to size. A tiny board that stacks on a Pi would take up less space than separate boards. Related to size is if the board keeps the camera port on the Raspberry Pi accessible.
Soldering
As you choose boards for a robot, you will note that some come as kits themselves, requiring parts to be soldered on. If you are already experienced with this, it may be an option. For experienced builders, this becomes a small cost in time depending on the complexity of the soldering. A small header is going to be a very quick and easy job, and a board that comes as a bag of components with a bare board will be a chunk of an evening.
Here, I will recommend components that require the least soldering.
Connectors
Closely related to soldering are the connectors for the motors and batteries. I tend to prefer the screw type connectors. Other types may require matching motors or crimping skills:
Conclusion
Our robot is space constrained; for this reason, we will be looking at the Raspberry Pi hat type form factor. We are also looking to keep the number of pins it binds to really low. An I2C-based hat will let us do this. The Full Function Stepper Motor Hat (also known as the Full Function Robot Expansion Board) gets us access to all the Pi pins while being a powerful motor controller:
It’s available in most countries, has space for the ribbon for the camera, and controls servo motors. I recommend the 4tronix PiConZero hat, or assembling a stack of PiBorg hats. These may be harder to source outside of the UK. The reader will need to adapt the code, and consider a tiny shim to retain access to the GPIO pins if using a different board.
In this article, we learned about selecting the parts needed to build a basic robot. We looked at the size, wheel, cost, and connectors for the robot chassis and a controller. To learn more about robotics and build your own robot check out this book Learn Robotics Programming.
Read next
Real-time motion planning for robots made faster and efficient with RapidPlan processor
Boston Dynamics adds military-grade mortor (parkour) skills to its popular humanoid Atlas Robot
Sex robots, artificial intelligence, and ethics: How desire shapes and is shaped by algorithms