Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Predicting Overtakes in Trucks Using CAN Data

Talha Hanif Butt, Prayag Tiwari, Fernando Alonso-Fernandez

TL;DR

The paper addresses predicting truck overtakes from real CAN-bus data to support ADAS safety. It evaluates three classifiers—Artificial Neural Networks, Random Forest, and Support Vector Machines (linear and RBF)—using a 1 s sliding window on CAN signals collected around an overtaking trigger, with data from three Volvo trucks. Results show high recall for overtakes ($TPR > 93\%$) but weaker no-overtake accuracy ($TNR$ typically $80\%$–$90\%$); combining Random Forest and linear SVM by averaging scores improves $TNR$ to $>92\%$ while maintaining $TPR$ near $91\%$ near the trigger, yielding a more balanced performance. This demonstrates the feasibility of CAN-based overtaking detection in trucks and suggests classifier fusion as a practical approach to achieve robust, balanced detection; future work includes optimizing window size and lead time, expanding the dataset, and exploring pseudo-labeled data and LSTM models for further gains.

Abstract

Safe overtakes in trucks are crucial to prevent accidents, reduce congestion, and ensure efficient traffic flow, making early prediction essential for timely and informed driving decisions. Accordingly, we investigate the detection of truck overtakes from CAN data. Three classifiers, Artificial Neural Networks (ANN), Random Forest, and Support Vector Machines (SVM), are employed for the task. Our analysis covers up to 10 seconds before the overtaking event, using an overlapping sliding window of 1 second to extract CAN features. We observe that the prediction scores of the overtake class tend to increase as we approach the overtake trigger, while the no-overtake class remain stable or oscillates depending on the classifier. Thus, the best accuracy is achieved when approaching the trigger, making early overtaking prediction challenging. The classifiers show good accuracy in classifying overtakes (Recall/TPR > 93%), but accuracy is suboptimal in classifying no-overtakes (TNR typically 80-90% and below 60% for one SVM variant). We further combine two classifiers (Random Forest and linear SVM) by averaging their output scores. The fusion is observed to improve no-overtake classification (TNR > 92%) at the expense of reducing overtake accuracy (TPR). However, the latter is kept above 91% near the overtake trigger. Therefore, the fusion balances TPR and TNR, providing more consistent performance than individual classifiers.

Predicting Overtakes in Trucks Using CAN Data

TL;DR

The paper addresses predicting truck overtakes from real CAN-bus data to support ADAS safety. It evaluates three classifiers—Artificial Neural Networks, Random Forest, and Support Vector Machines (linear and RBF)—using a 1 s sliding window on CAN signals collected around an overtaking trigger, with data from three Volvo trucks. Results show high recall for overtakes () but weaker no-overtake accuracy ( typically –); combining Random Forest and linear SVM by averaging scores improves to while maintaining near near the trigger, yielding a more balanced performance. This demonstrates the feasibility of CAN-based overtaking detection in trucks and suggests classifier fusion as a practical approach to achieve robust, balanced detection; future work includes optimizing window size and lead time, expanding the dataset, and exploring pseudo-labeled data and LSTM models for further gains.

Abstract

Safe overtakes in trucks are crucial to prevent accidents, reduce congestion, and ensure efficient traffic flow, making early prediction essential for timely and informed driving decisions. Accordingly, we investigate the detection of truck overtakes from CAN data. Three classifiers, Artificial Neural Networks (ANN), Random Forest, and Support Vector Machines (SVM), are employed for the task. Our analysis covers up to 10 seconds before the overtaking event, using an overlapping sliding window of 1 second to extract CAN features. We observe that the prediction scores of the overtake class tend to increase as we approach the overtake trigger, while the no-overtake class remain stable or oscillates depending on the classifier. Thus, the best accuracy is achieved when approaching the trigger, making early overtaking prediction challenging. The classifiers show good accuracy in classifying overtakes (Recall/TPR > 93%), but accuracy is suboptimal in classifying no-overtakes (TNR typically 80-90% and below 60% for one SVM variant). We further combine two classifiers (Random Forest and linear SVM) by averaging their output scores. The fusion is observed to improve no-overtake classification (TNR > 92%) at the expense of reducing overtake accuracy (TPR). However, the latter is kept above 91% near the overtake trigger. Therefore, the fusion balances TPR and TNR, providing more consistent performance than individual classifiers.
Paper Structure (6 sections, 5 equations, 4 figures, 4 tables)

This paper contains 6 sections, 5 equations, 4 figures, 4 tables.

Table of Contents

  1. INTRODUCTION
  2. Experimental framework
  3. Database
  4. Classifiers
  5. Results
  6. Conclusions

Figures (4)

  • Figure 1: Boxplot of scores towards class0 (left column, no overtake class) and class1 (right, overtake) from -10 to +1 seconds around the trigger. From top to bottom row: ANN, RF, SVM linear and SVM rbf classifiers.
  • Figure 2: Precision-Recall curves of the classifiers at different moments before the overtake maneuver starts. AUC (Area under the curve) values are given in Table \ref{['tab1']}.
  • Figure 3: $F1$-score vs. threshold at different moments before the overtake maneuver starts.
  • Figure 4: Graphical plot of $TPR/TNR$ at different moments before the overtake manoeuvre starts ($t$ corresponds to the precondition trigger, $t$-1 to one second earlier, and so on).