Spatiotemporal Contextual Analysis of Driving Behavior to Ensure Road Safety
As the human population continues to grow, generating more traffic, the need for timely travel is crucial for various workplace commitments. However, this surge in activity also amplifies concerns about traffic safety, particularly related to driving behavior. Often reckless driving or the lack of skills among novice drivers contribute to accidents on the roads. In light of such concerns, we look into the problems arising due to driving behavior and how we can utilize its characteristics in enhancing the traffic safety to a next level. First, we look into the problem of driving behavior that are often influenced by the actions of surrounding vehicles. We develop a smartphone based pervasive sensing system that utilizes IMU & Video data to investigate the causes and consequences of driving maneuvers to score a driver based on a thorough understanding of their on-road driving behavior. Next, we inspect whether a sudden fluctuation in driving behavior is due to either a lack of driving skill or the effect of various on-road spatial factors such as pedestrian movements, peer vehicles’ actions, etc. Consequently, we look into the opportunities in understanding the safety of individual road junctions utilizing such contextual information responsible for poor driving. We develop a system that automatically annotates the road segments with a driving safety level to aid cautious maneuvering and safe driving practices. Finally, we focus on individual driving behaviors and their impact on the overall traffic dynamics of a smart city. In a nutshell, this study provides a crucial basis for exploring city traffic dynamics, highlighting the importance of understanding how driving behavior relates to traffic management.
Key Features
Uniqueness: The work associated with this thesis annotates the driving behavior along with explainable causal contexts from driving signatures under the influence of diverse environmental conditions. The study also predicts traffic incidents and annotates risk factors on a digital maps through the propagated effect of inferred contexts.
- Multimodal: The framework uses different modalities such as IMU, GPS and camera to develop the work in 4 parts.
- Automated Annotation: One of the novel contributions is to annotate driving behavior, risk factors associated with road junctions over a digital map and mark the causal contexts just-in-time pervasively.
- Reduction of Human Effort: The main stakeholders like drivers, passengers, traffic polices can see the annotations in a smartphone app with interactive features and least distractions during driving. The system drastically cuts the energy consumptions (low battery usage) and involvement of costly sensors. A smartphone is enough to provide the solutions handy.
- Human annotations: VR platform was designed to collect human annotations for initial development of the framework.
- Real-world Scenarios: All the parts of the framework are developed with real-world data, where two parts DriCon and DriveR are deployed over a real-world driving environment with more than 20 users with detailed usability study to get user feedback on the field. Rest of the parts were tested in simulated environments.
- Deeper Insights on Traffic Environment: Finally, we get valuable insights about the collaborative tasks driving and how the inter-relationship of drivers and surrounding environment influence each other. The motto of the framework is to ensure road safety by maximizing the effort of letting the drivers know about their shortcomings on the go and also proactively probing them about any anomalous traffic incidents and associated risk factors ahead of their travel.
Contributors
Debasree Das
IIT Kharagpur, India
Sajal K. Das
Missouri University of S&T
Bivas Mitra
IIT Kharagpur, India
Sandip Chakraborty
IIT Kharagpur, India
Teaser Video
Publications
- Debasree Das, Sandip Chakraborty, and Bivas Mitra. "DriveR: Towards Generating a Dynamic Road Safety Map with Causal Contexts." Proceedings of the ACM on Human-Computer Interaction 8, no. MHCI (2024): 1-35.
- Debasree Das, Shameek Bhattacharjee, Sandip Chakraborty, Bivas Mitra, and Sajal K. Das. "Early Detection of Driving Maneuvers for Proactive Congestion Prevention." In 2024 IEEE International Conference on Pervasive Computing and Communications (PerCom), pp. 135-142. IEEE, 2024.
- Debasree Das, Sandip Chakraborty, and Bivas Mitra. "DriCon: On-device just-in-time context characterization for unexpected driving events." In 2023 IEEE International Conference on Pervasive Computing and Communications (PerCom), pp. 12-21. IEEE, 2023.
- Debasree Das, Sugandh Pargal, Sandip Chakraborty, and Bivas Mitra. "Dribe: on-road mobile telemetry for locality-neutral driving behavior annotation." In 2022 23rd IEEE International Conference on Mobile Data Management (MDM), pp. 159-168. IEEE, 2022.
- Debasree Das, Sugandh Pargal, Sandip Chakraborty, and Bivas Mitra. "Why slammed the brakes on? auto-annotating driving behaviors from adaptive causal modeling." In 2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops), pp. 587-592. IEEE, 2022.
- Debasree Das, Pragma Kar, Sugandh Pargal, and Sandip Chakraborty. "FreeSteer: A Smartphone Application for Detecting Anxiety in Novice Drivers through Smart Glasses." In 2023 15th International Conference on COMmunication Systems & NETworkS (COMSNETS), pp. 427-431. IEEE, 2023.
- Rohit Verma, Sugandh Pargal, Debasree Das, Tanusree Parbat, Sai Shankar Kambalapalli, Bivas Mitra, and Sandip Chakraborty. "Impact of Driving Behavior on Commuter’s Comfort During Cab Rides: Towards a New Perspective of Driver Rating." ACM Transactions on Intelligent Systems and Technology (TIST) 13, no. 6 (2022): 1-25.