Combining locating systems for a more accurate positioning
Combining locating systems for a more accurate positioning
The media often talk about autonomous vehicles. However, it is obvious that driverless cars won’t take over our streets anytime soon. What is talked about a lot less, is that in industrial environments, autonomous applications have been integrated a lot more already. AGVs or Automated Guided Vehicles can execute different tasks in a production environment: transport, stock-taking, locating, etc. For this, they need to know where they are, where they want to go and what is the best route – taking into account other static or moving obstacles on the way.
The design and calibration of the positioning systems of those AGVs is therefore crucial. And certainly not obvious. Indeed, AGVs are usually deployed for indoor applications. GPS tracking, which would work outdoors, is not sufficient indoors. The tracking systems have three important functions:
- Keeping the vehicle on the right track, taking into account the environment, other (autonomous) vehicles and objects and/or persons.
- Establishing the most efficient route and distributing the tasks between the different vehicles in a fleet.
- Preventing collisions with other vehicles, objects and/or persons.
Slow, expensive and inaccurate
Unfortunately, the existing methods on which positioning systems are based do not meet industrial requirements for accuracy, update speed, robustness and cost. Therefore, Flanders Make partnered up with industrial businesses such as Bekaert and Octinion to develop an innovative approach combining different existing positioning techniques, i.e.:
- UWB (ultra-wide-band) sensors
- Accelerometers and gyroscopes
- Visual positioning using cameras
Depending on the application and preconditions such as cost, environment, scalability, etc., several combinations are possible. Therefore, a design framework was developed that makes it possible to carry out a quantitative comparison of the techniques in a relatively simple way. This enables developers to rapidly simulate the impact of adding/removing a certain technique or of adapting a series of parameters. As a result, the optimal combination can be established a lot faster.
Large-scale deployment of AGVs
In the framework of an ICON-project, we developed - together with several companies - technology for AGVs. Bekaert and Octinion, amongst others, are currently using this technology for their own applications
Bekaert is a Belgian multinational business, specialised in steel wire, steel rope and advanced materials and coatings. The company uses Decawave semiconductors for the positioning of AGVs on its production site. Such an industrial environment is far from ideal for accurate positioning, e.g. due to flying dust particles. Together with Flanders Make, specific software (firmware) has been developed for the Decawave chips that are used for processing the signals that are required for sensor data fusion. This is a combination of data from different sensors in order to achieve the most accurate measurement possible. By optimising the use of the redundant information from all sensors, positioning is more accurate and the system can filter out reflection. Thanks to sensor data fusion, Bekaert can further improve and upscale its autonomous navigation systems.
Octinion is an innovative R&D business specialising in mechatronic product development and automation solutions. They developed an indoor positioning system for ‘agrobots’. For this, they had to map out why autonomous indoor navigation sometimes fails. This can be caused, e.g., by the feedback from the controller, the actual sensors or the chosen combination of sensors. Together with Flanders Make, a prototype (diagnostics tool) has been developed that assesses the impact of the environment around the AGV on the UWB measuring quality. With this tool, Octinion can establish the quality of the signals with great accuracy (reach, signal strength), which considerably speeds up the development process of new autonomous applications