Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Driver Fatigue Prediction using Randomly Activated Neural Networks for Smart Ridesharing Platforms

Sree Pooja Akula, Mukund Telukunta, Venkata Sriram Siddhardh Nadendla

TL;DR

This work addresses predicting driver stopping decisions in ridesharing by modeling a Dynamic Discounted Satisficing (DDS) heuristic that captures fatigue- and time-driven changes in decision thresholds. It introduces a stochastic neural network with random activations that implements DDS and a Sampling-Based Back Propagation Through Time (SBPTT) training algorithm to learn DDS parameters from sequential data, validated on both simulated scenarios and the Chicago taxi dataset. The results show DDS-based models outperform Discounted Satisficing baselines, with improvements that scale with the number of random samples $R$, suggesting better tracking of fatigue dynamics and decision making. The approach offers practical implications for ridesharing platforms to optimize incentives and ride-offers by anticipating drivers’ stopping behavior under fatigue, thereby improving overall efficiency and revenue.

Abstract

Drivers in ridesharing platforms exhibit cognitive atrophy and fatigue as they accept ride offers along the day, which can have a significant impact on the overall efficiency of the ridesharing platform. In contrast to the current literature which focuses primarily on modeling and learning driver's preferences across different ride offers, this paper proposes a novel Dynamic Discounted Satisficing (DDS) heuristic to model and predict driver's sequential ride decisions during a given shift. Based on DDS heuristic, a novel stochastic neural network with random activations is proposed to model DDS heuristic and predict the final decision made by a given driver. The presence of random activations in the network necessitated the development of a novel training algorithm called Sampling-Based Back Propagation Through Time (SBPTT), where gradients are computed for independent instances of neural networks (obtained via sampling the distribution of activation threshold) and aggregated to update the network parameters. Using both simulation experiments as well as on real Chicago taxi dataset, this paper demonstrates the improved performance of the proposed approach, when compared to state-of-the-art methods.

Driver Fatigue Prediction using Randomly Activated Neural Networks for Smart Ridesharing Platforms

TL;DR

This work addresses predicting driver stopping decisions in ridesharing by modeling a Dynamic Discounted Satisficing (DDS) heuristic that captures fatigue- and time-driven changes in decision thresholds. It introduces a stochastic neural network with random activations that implements DDS and a Sampling-Based Back Propagation Through Time (SBPTT) training algorithm to learn DDS parameters from sequential data, validated on both simulated scenarios and the Chicago taxi dataset. The results show DDS-based models outperform Discounted Satisficing baselines, with improvements that scale with the number of random samples , suggesting better tracking of fatigue dynamics and decision making. The approach offers practical implications for ridesharing platforms to optimize incentives and ride-offers by anticipating drivers’ stopping behavior under fatigue, thereby improving overall efficiency and revenue.

Abstract

Drivers in ridesharing platforms exhibit cognitive atrophy and fatigue as they accept ride offers along the day, which can have a significant impact on the overall efficiency of the ridesharing platform. In contrast to the current literature which focuses primarily on modeling and learning driver's preferences across different ride offers, this paper proposes a novel Dynamic Discounted Satisficing (DDS) heuristic to model and predict driver's sequential ride decisions during a given shift. Based on DDS heuristic, a novel stochastic neural network with random activations is proposed to model DDS heuristic and predict the final decision made by a given driver. The presence of random activations in the network necessitated the development of a novel training algorithm called Sampling-Based Back Propagation Through Time (SBPTT), where gradients are computed for independent instances of neural networks (obtained via sampling the distribution of activation threshold) and aggregated to update the network parameters. Using both simulation experiments as well as on real Chicago taxi dataset, this paper demonstrates the improved performance of the proposed approach, when compared to state-of-the-art methods.
Paper Structure (15 sections, 6 equations, 9 figures, 1 algorithm)

This paper contains 15 sections, 6 equations, 9 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Modeling Driver's Cognitive Atrophy
  3. Dynamic Discounted Satisficing
  4. Illustrative Example
  5. Need for DDS Model
  6. Modeling DDS using Neural Networks
  7. Architecture Design
  8. Training Algorithm and Performance Metrics
  9. Experiment Design
  10. Simulation Experiment
  11. Chicago Taxi Dataset and Preprocessing
  12. Results and Discussion
  13. Simulation Results
  14. Results on Chicago Taxi Dataset
  15. Conclusion and Future Work

Figures (9)

  • Figure 1: Illustrative Example of Three Different Decision-Making Models over 5 Tasks.
  • Figure 2: Model Architecture
  • Figure 3: Task Decision Maker Architecture
  • Figure 4: Dynamic Threshold Architecture
  • Figure 5: Dynamic Threshold Architecture
  • ...and 4 more figures

Theorems & Definitions (1)

  • Definition 1