The RUMBy (ex DUMBo) Project
RUMBy (Reduced-scale UnManned BuggY) is an internal project of the Mechatronics Lab aimed at the developement of an autonomous vehicle able to drive in dynamic conditions (i.e. high speed).
The peculiarity of the project is the adoption of a radiocontrolled amateur 1:6 scale buggy. The relatively low power of the vehicle (4 hp) requires to keep the weight and the power consumption of the on-board electronics as low as possible, thus we decided to have a ground-based computer doing all the computational-intensive tasks (path planning and state monitoring) and controlling the buggy via the RC joystick.
Architecture
At the moment, the baseline towards the RUMBy (ex DUMBO) autonomous vehicle is as described in the previous post.
In a few words, the idea is to have a manager computer (running OS X) that controls the vehicle sending throttle and steering commands through a D/A converter (NI USB-6008) connected with a standard RC digital joystick. The control loop is closed by sensor fusion data transmitted by the on-board PC104 via 802.11n gateway.
The lower part of the scheme depicts the software running on the manager computer. It is a state machine developed in Ruby thanks to the Statemachine library, which helps in keeping the system modular and easily extendable. The various machine states can load different low-level control and path planning modules written in C/C++ and dynamically linked with the Ruby interpreter. This hybrid solution provides fast computation and easy high-level management at the same time.
Breadboarding and Prototyping
To let the computer remotely drive a 25 kg vehicle at 50 km/h is a risky business.
For that reason, we are currently linking the manager computer with a cheaper, slower, and lighter RC model powered by an electric motor (the one shown in the video). Given the reduced scale of this vehicle, we also have to reduce the number of onboard sensors: one three-channels Bluetooth module sending longitudinal and lateral accelerations and motor RPMs.
The first experiment is the one shown in the video: the steering and throttle control signals have been linked with the acceleration signals measured by the two axes Bluetooth accelerometer. This resulted in a kind of wireless joystick with a surprisingly easy and intuitive use.
The next step is to mount the accelerometers onboard of the vehicle and perform easy computer-controlled longitudinal acceleration maneuvers, in order to have a dynamic identification of the vehicle performance along the longitudinal axis.
The next figure shows the results of one of these tests, reporting the plots of longitudinal acceleration, speed, and engine tension. The vehicle speed is integrated by the acceleration signal. The sudden change in acceleration profile after 0.8 s is due to tire slippage.
