As the automatic driving vehicle is coming, it is difficult to develop its software and hardware art and crafts. This is not surprising: after all, this area is still in the starting stage. Design a powerful system involves the most advanced software and (possibly) risk and expensive hardware. Therefore, the entry bar is set very high, only companies with large resources, such as Alphabet's Waymo.
I believe that for those who want to try it, the approximation of this technology can be used, just like using VR and Google Cardboard. The project's focus is to build a low-cost DIY robot car with advanced sensors, drives, and connections. The goal is to allow anyone to implement cutting-edge robots and automatic driving algorithms on this small robot.
As a software engineer, I often think of democratization and openness in the field of my work. I often ignore the miracle of the most advanced technology, just disclose on Github, some terminal orders are far from working on my desktop.
Fortunately, the automatic driving car field seems to be a step toward this direction. There are people interested in this theme offer MOOC and free online information. However, they still lack hardware platforms, and learners can safely and quickly assess their ideas. The best hope now is the simulator. I believe that for those who want to try it, the approximation of this technology can be used, just like using VR and Google Cardboard.
Can you guess what is going on?
The platform I have proposed is different from Google Cardboard. The principle remains unchanged: low cost (or made by something you may have), you can start using it without expertise. I have been committed to the remote control car based on smartphones, which can reuse the sensors (GPS, compass, accelerometers, lights), connection and high resolution cameras on all modern devices.
design
There are 3 main components of the robot:
3D printer
The main body of the robot is basically an advanced smartphone shell.
Since the phone is placed, the camera will usually look at the ceiling, which is not very useful. Therefore, there is a mirror at the top of the camera, placed at a 45 degree angle, so that the robot can look forward.
Many active components are placed on the chassis. The car is driven by a brushless motor driven by a series of deceleration gears to the rear axle. The AR perfog encoder is connected to the axis of the motor. The steering is controlled by a small servo control located in front of the chassis.
Car electronics
In order to control the actuator and read telemetry data, a small amount of electronic components are installed on the chassis.
The main circuit board is based on excellent NRF52 SOC. It provides Bluetooth Le Radio and mobile phone communication. The servo is directly controlled by the chip, but the motor requires an additional electronic speed controller (ESC).
software
To simplify the development, the software is divided into computers and mobile apps. The goal is to let the app as simple as possible, and only the most basic connection. This makes the application not in the iterative development cycle, due to the more powerful hardware and robotic software availability (such as ROS), this is much faster on your computer. Disadvantages is that robots must always be connected to the computer (via WiFi) and some delays are introduced when reading the sensor.
Due to the video processing and connection function of modern smartphones, delayed to the lowest - camera image delay 80ms, transmission steering commands and receive local WiFi telemetry. Delay. The robot can be controlled from a comfortable 10 Hz + loop by entrusting real-time control of the vehicle electronic device.
Thanks to various available sensors, the car can be used to achieve a variety of robots and self-driven algorithms - the powerful positioning of the accelerometer, the compass, encoder, and GPS data is combined to use computer vision (or machine learning).
Be
Article source network
Our other product:
