3 steps to embed autonomous navigation in unmanned vehicles – and 1 shortcut
Designing autonomous navigation for an unmanned vehicle is a process that requires forward thinking and making the right choices. There are however just a few steps to follow. This article describes these steps and gives a useful tip for a development shortcut.
Step 1: Determining the conditions
The first step is to determine where the navigation system will be used. Is there GNSS-coverage (GPS, GLONASS, BeiDou)? How long are the GNSS outages? A holonomic constraint is the assumption that the direction of the vehicle is also the direction of the GPS-course, can this assumption be applied? Will there be vibrations? Is heading sufficient and is velocity determined by e.g. wheel speed or is a full navigation system required? Can the magnetic field be used for heading and what will be the intensity of the vibrations?
All these environmental conditions and characteristics of the vehicle greatly influence the choice of the sensor systems.
Step 2: Choosing and characterizing the sensor systems
The second step is the selection of the sensors. Wheel speed sensors and laser scanners are used in autonomous navigation. At the heart of the system architecture, there is an Inertial Measurement Unit (IMU), Attitude and Heading Reference System (AHRS) or even an Inertial Navigation System (INS) with GPS/GNSS. A GNSS receiver can also be chosen separately from an AHRS, especially when the application is relatively benign. Data can be used in algorithms for ADAS systems, correction algorithms for camera images and of course as a source for position and velocity data.
Accuracy and costs often go hand in hand and that is the reason that Xsens has a wide range of products that have the same software interface. This means that there is always a choice that fits the budget and meets the requirements. The integration level also influences the costs: AHRSs are higher priced than IMUs as they are turnkey products.
Next to accuracy, it is important to understand for which purpose IMUs, AHRSs and GNSS/INSs are built. IMUs in the lower range are often designed for the wearable market, where vibrations are non-existent. Using these type of sensors can only be done when filtering, calibration and proper mounting have been taken care of. AHRSs and GNSS/INSs with higher grade sensors are excellent for autonomous navigation, though it is still good to check whether different settings can be applied for different situations and/or how well they can cooperate with other sensor systems.
Step 3: implementing the system architecture
The last step before testing is to actually develop the autonomous navigation functionality. This is definitely not a straightforward task, and often it is complicated by rigid communication to and from the IMU or AHRS. A well-documented protocol is preferable as well as sample code and drivers on e.g. ARM® mbedTM and ROS. Since large semiconductor companies cannot support individual customers that use a limited number of sensors, it may be worth investigating the use of value-added resellers or system integrators in order to reach a shorter integration time.
Once you have your configuration ready, your car is ready to drive!
The easiest way to get autonomous navigation is to find one of the specialized companies that have experience in providing autonomous technology solutions. One of these companies in the US is AutonomouStuff, the world leader supplying components and services that enable autonomy. Entire kits provide a one-stop solution for autonomous navigation. The kits can save extensive development time, enabling a fast track to testing and deployment. If you like to know more about AutonomouStuff, read the customer case we published.