With mobile robotics being applied for more and more complex applications, their autonomy should be preserved. While a lot of research is performed into the direction of failure prediction for autonomous processes or systems, the field of mobile robots has received less attention. Proactive failure prediction for mobile robots is a useful tool to prevent unwanted downtime and undesired damages. This work attempts to fill this research gap by showing the applicability of anomaly detection methods for failure prediction in the field of mobile robots. Specifically, we employ an unsupervised Variational Autoencoder to predict failures in the operational data from the Discovery Collector, a manure cleaning robot developed by Lely Industries. We elaborately showcase the feature engineering steps which yield the best performance, provide the performance of three general datasets, and state promising next steps for root cause classification which is enabled by accurate failure prediction. All in all, our work shows that the use of feature offsets, calculated from desired values compared to actual values, enhances the model performance tremendously. The provided datasets showcase F1-scores ranging from 0.64-0.76, showing the proposed solution is able to solve the failure prediction problem in the field of mobile robots, while highlighting the encountered limitations for future improvement.

Fuzzing in Big Data applications is a relatively new field which is still lacking effective tools to support automated testing. Recently, a framework called BigFuzz was published which made fuzz testing for big data systems feasible. But there was no solution to work with Big Data programs that use JSON typed data. Big Data systems often make use of JSON typed data and JSON typed fuzzers for Big Data systems are currently not publicly found. With this work it is now possible to support JSON typed input data and apply fuzzing per iteration. The work requires a user defined input specification of the set of valid JSON inputs for the program under test, and a converted Java program based on the Spark program to test. However, it is almost certain the latter is not necessary in the future since it is likely this conversion can be automated.

This work is shown to be effective in finding bugs in a rather small amount of trials. Oppositely, it loses the descriptive exceptions, since it finds bugs later in the program instead of at the input validation phase. The work still has its limits to be applied extensively in the field of automatic testing, but serves as a proof of concept that automatically finding bugs in Big Data applications working with JSON typed data is in fact possible.