Conference Papers
Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506
Browse
3 results
Search Results
Item Driver Skill Profiling Using Machine Learning(Springer Science and Business Media B.V., 2024) Akhtar, N.; Mohan, M.Road safety is a critical aspect of public safety, and driving skills are essential to ensuring safety on the road. An accurate understanding of one's driving abilities is crucial in promoting safe driving practices and reducing the risk of accidents. Overconfidence and underestimating road events can lead to a false sense of handling emergencies and may result in a higher risk of traffic offenses and accidents. Young and novice drivers are particularly susceptible to these issues and may overestimate their abilities, leading to a higher risk tolerance. Machine learning is a viable approach that can compare perceived and actual skills to measure subjective driving skills accurately. A scoring system based on machine learning algorithms can quantify driver skills effectively and improve self-awareness, ultimately contributing to increased road safety. The proposed scoring system can give drivers an accurate assessment of their abilities, helping them take necessary corrective actions to work on their weaknesses. Driving style, encompassing violations, errors, and lapses, and driving skills, including perceptual motor skills and safety skills, are the two main components of the human factor in driving. Training sessions may be conducted based on the proposed scoring system using machine learning that can help improve drivers' self-awareness and reduce the risk of accidents. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.Item Classification of Longitudinal Driving Events Using Vehicle Response Signals for Profiling Driving Behaviors(Springer Science and Business Media Deutschland GmbH, 2025) Arichandran, R.; Krishnakar, H.S.; Kumar, A.; Mohan, M.Driving behavior profiling (DBP) involves evaluating driving patterns to determine a safety score for drivers. Proper classification of driving events increases the accuracy of profiling driving behaviors. Most driving event classification models consider lateral driving events, such as turning and lane changes. In heavy traffic conditions, it is impossible to perform lateral events independent of the position of other vehicles. This research aims to develop a model for classifying longitudinal driving events (acceleration and braking) and nonevents using vehicle response signals. To develop the model, naturalistic driving data were collected using a passenger car on a 19 km road stretch. Vehicle response signals were collected using Inertial Measurement Unit (IMU) sensors fixed on the test vehicle with a frequency of approximately 200 Hz with timestamps. The driver’s pedal operation was also captured with timestamps using a camera to map the ground truth labels with vehicle response signals. The data were collected from 5 drivers, totaling a dataset for approximately 190 km. The start and end times of all 634 events (444 driving events and 190 nonevents) were used to label the driving events in the IMU sensor data. These labeled driving events were split into the train (476 events) and test (158 events) datasets. Hidden Markov Model (HMM) algorithm was used to develop classification models for the driving events. The models were developed for various combinations of accelerations using the training dataset. The accuracy of these models was then compared to a test dataset. The models achieved 90.99% and 77.08% accuracy, respectively, in classifying events and nonevents using data from the accelerometer. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.Item Classification of Lateral Driving Events Using Vehicle Response Signals for Profiling Driving Behaviors(Springer Science and Business Media Deutschland GmbH, 2025) Arichandran, R.; Kumar, A.; Krishnakar, H.S.; Mohan, M.Classification of driving events is crucial in profiling driving behaviors, which could significantly increase road safety. In several studies classifying driving events, drivers were asked to perform hard turns and lane changes. However, these represented simulated situations much different from real-life scenarios. The main aim of this study was to classify lateral driving events (turn) and non-events from naturalistic driving data in actual driving conditions. A stretch of 8 km state highway was identified as the study road, and the data were collected using 8 drivers. The acceleration and gyroscope data were collected using Inertial Measurement Unit (IMU) sensors with a frequency of approximately 200 Hz with the timestamps. A dashboard camera was fixed to capture the driver’s view with timestamps. The start and end times of the turns (left turn and right turn) and non-events were manually marked using the timestamps in the recorded videos. The total count of marked events and non-events was 1246, and their start and end times were used to label the driving events in the IMU sensor data. These labeled driving events were split into the train (934 driving events) and test (312 driving events) datasets. The Hidden Markov Model (HMM) algorithm was adopted to create classification models for the driving events. HMM models were developed using the training dataset for various features, such as lateral and longitudinal acceleration. The accuracy of these models was then compared to a test dataset. The models achieved 96.09% and 95.1% accuracy in classifying turns and non-events using data from the gyroscope’s y-axis. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.