Browsing by Author "Saaristola, Tomas"

Now showing 1 - 2 of 2

Data collection system for autonomous vehicles
(2022-05-16) Saaristola, Tomas
Sähkötekniikan korkeakoulu | Master's thesis
Recently, an increasing number of public transport services have begun incorporating autonomous driving technology. Among other benefits, autonomous vehicles are utilized due to their ability to improve traffic and make public transport more accessible. The operation and development of autonomous vehicles requires collecting the data recorded during driving. Due to the numerous sensors and complex software inherent to autonomous vehicles, the rate of data generation is massive. Although data collection is a challenging problem, few previous works have addressed the issues in detail. The aim of this thesis was to design, implement and validate a data collection system for autonomous vehicles. Legislation, standards and the current data collection process at Sensible 4 were considered for the design of the system. To target all areas of the data collection process, the system design was divided into three components: data selection, data extraction, and data transmission. The system was validated by implementing an extensive testing suite. All three components of the system performed according to the design requirements. The selection component selected data to be collected based on predefined criteria. The extraction component extracted the data samples from recordings and compressed them. The transmission component transmitted the data samples while minimizing disruption to other outgoing connections. The validation confirmed that together the system components can collect data samples from an online autonomous vehicle. This thesis introduced a detailed solution to the data collection problem of autonomous vehicles. Although covering the whole problem area, some aspects of the system should be further developed and tested to achieve a robust data collection solution. Future work could address the performance of the transmission component in unreliable mobile networks and implement more sophisticated data selection methods to cover different types of data collection events.
Proportional estimation of finger motions
(2019-05-20) Saaristola, Tomas
Sähkötekniikan korkeakoulu | Bachelor's thesis