Since few month we started to design a wheeled robot base on the android OS. We are a group of robot’s addicted, professional or not, and we share our off time doing robot as an hobby.
We found many wheeled robot on the market, but nothing corresponding to our goals. So, we decide to design the one we want. On a previous post, Bilock introduced the Scalpel project. Here is more details about the Scapel’s mechanic design.
Our robot must be :
- Android based
- Fast, precise and responsive
- Fully autonomous
- Robust and easily assembled
We try to keep the design modular, so you could use this robot locomotion in your own project.
For each topic introduced here, I tried to give a theoretical overview combined with some manufacturing details. I would also try to give you tip and trick to create your own robot based on this scalpel kit. Later, with the scalpel team, we will take a confrontation between the theoretical analysis and experimental results.
Capacities of responsive and accurate displacements
The first step is to determine the mechanical characteristics of the robot’s locomotion. The reactivity of the platform is directly connected to the acceleration and deceleration capabilities without loss of grip. We will see the importance of choosing wisely the coordinates of the center of gravity and the importance of the grip between wheels and the ground.
The electric motors
Then we explain the motor choice to achieve the desired performance. We will detail the operating characteristics of a DC motors and the importance of specific power supplies. Electric motors for our robot, are controlled by PWM command. We will describe the signal command characteristics to optimise the locomotion precision and reactivity while securing the hardware against possible damage.
The control systems of rotary sensor wheels
Odometric wheels are use to estimate the position of a moving robot. This system is generally made by wheels connected to rotary encoders. Encoder are generally directly linked to the motor, but this assumes that the wheel is permanently in the contact with the ground and the friction is perfect. Here we explore an original solution of a decoupled odometric wheels. Encoders that are available on the market are expensive and the mechanical integration of this decoupled wheel is not trivial. I would show how, using an Hall effect sensors “1024 increments per revolution” and the minimum number of mechanic parts, that it is possible to manufacture a robust encoder wheel for an affordable cost.
Systems of perceptions by IR sensors
Since we want this robot being autonomous, we need to use efficient obstacle detection sensors in order to avoid obstacle collision. As a first step, we integrated the famous Sharp GP2D120. I know that software developers are never happy, they always want to modify the IR sensors angle but the Sharp senors packaging do not allow rapid orientation of the sensors. I will detail the mechanical changes I did on the Sharp infrared sensors packaging to allow my “never satisfied” friends to change quickly the sensor orientation.