Uncertainty-aware weather-related road surface condition classification

Sammanfattning

Autonomous vehicles rely on collaborative systems for decision-making during operation,one of which is machine learning algorithms. Using cameras and sensorsmounted on the car, these algorithms can analyse captured data to interpret surroundings,enabling it to make informed and safe driving decisions. The algorithmsare trained on data gathered in the field. In real-time use, samples that do not belongto the training distribution, referred to as Out-Of-Distribution (OOD) samples,can occur, which the algorithm then confidently uses to make driving decisions. Ina worst-case scenario, this can lead to injuring passengers or property. The primaryobjective of this thesis is to implement and analyse methods for detecting OODsamples in a road surface condition model. The machine learning algorithm was aconvolutional neural network, trained to detect dry, wet and snowy road surface conditions.The methods used to detect samples were Maximum Softmax Probability,Energy-Based OOD detection, Outlier Exposure, Virtual Outlier Synthesis, RectifiedActivations, Deep Nearest Neighbour, and Virtual Logit Matching. Evaluationof the methods involved three datasets - Cars, Slush and Glare - constructed fromthe same database as the training data, thus considered as near OOD datasets. Additionally,two publicly available datasets, CIFAR-10 and Texture were used as farOOD datasets. Deep Nearest Neighbour and Virtual Logit Matching performed thebest, achieving near-perfect results, when evaluated on the far OOD datasets. Whentested on the cars dataset, Virtual Logit Matching exhibited a notable deviation inperformance compared to the other methods, although it still performed poorly. Onthe rest of the near OOD data, the detectors did not outperform the two baselinedetectors, Energy-Based OOD detection and Maximum Softmax Probability.