Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Using Machine Learning to Take Stay-or-Go Decisions in Data-driven Drone Missions

Giorgos Polychronis, Foivos Pournaropoulos, Christos D. Antonopoulos, Spyros Lalis

TL;DR

The paper tackles minimizing mission time in data-driven drone missions by learning stay-or-go decisions at points of interest under time-varying event probabilities. It introduces a perceptron-based predictor and reinforcement learning approaches (DQN, A2C, PPO) and evaluates them against a regression baseline, showing substantial improvements in worst-case time and near parity with an oracle under various dynamics. The results demonstrate strong robustness and adaptability, with ML methods outperforming regression across patterns A and B, especially as probability changes become abrupt. The findings underscore the practical potential of ML-driven runtime decisions to accelerate drone missions without sacrificing correctness, and suggest avenues for richer decision-making and faster retraining in future work.

Abstract

Drones are becoming indispensable in many application domains. In data-driven missions, besides sensing, the drone must process the collected data at runtime to decide whether additional action must be taken on the spot, before moving to the next point of interest. If processing does not reveal an event or situation that requires such an action, the drone has waited in vain instead of moving to the next point. If, however, the drone starts moving to the next point and it turns out that a follow-up action is needed at the previous point, it must spend time to fly-back. To take this decision, we propose different machine-learning methods based on branch prediction and reinforcement learning. We evaluate these methods for a wide range of scenarios where the probability of event occurrence changes with time. Our results show that the proposed methods consistently outperform the regression-based method proposed in the literature and can significantly improve the worst-case mission time by up to 4.1x. Also, the achieved median mission time is very close, merely up to 2.7% higher, to that of a method with perfect knowledge of the current underlying event probability at each point of interest.

Using Machine Learning to Take Stay-or-Go Decisions in Data-driven Drone Missions

TL;DR

The paper tackles minimizing mission time in data-driven drone missions by learning stay-or-go decisions at points of interest under time-varying event probabilities. It introduces a perceptron-based predictor and reinforcement learning approaches (DQN, A2C, PPO) and evaluates them against a regression baseline, showing substantial improvements in worst-case time and near parity with an oracle under various dynamics. The results demonstrate strong robustness and adaptability, with ML methods outperforming regression across patterns A and B, especially as probability changes become abrupt. The findings underscore the practical potential of ML-driven runtime decisions to accelerate drone missions without sacrificing correctness, and suggest avenues for richer decision-making and faster retraining in future work.

Abstract

Drones are becoming indispensable in many application domains. In data-driven missions, besides sensing, the drone must process the collected data at runtime to decide whether additional action must be taken on the spot, before moving to the next point of interest. If processing does not reveal an event or situation that requires such an action, the drone has waited in vain instead of moving to the next point. If, however, the drone starts moving to the next point and it turns out that a follow-up action is needed at the previous point, it must spend time to fly-back. To take this decision, we propose different machine-learning methods based on branch prediction and reinforcement learning. We evaluate these methods for a wide range of scenarios where the probability of event occurrence changes with time. Our results show that the proposed methods consistently outperform the regression-based method proposed in the literature and can significantly improve the worst-case mission time by up to 4.1x. Also, the achieved median mission time is very close, merely up to 2.7% higher, to that of a method with perfect knowledge of the current underlying event probability at each point of interest.

Paper Structure

This paper contains 18 sections, 5 equations, 3 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Formulation
  4. Mission Execution
  5. Perceptron Approach
  6. Reinforcement Learning Approach
  7. Training of RL agent
  8. State space, action space and reward function of RL agent
  9. Algorithms
  10. Evaluation
  11. Topology and change of event probability
  12. Fine-tuning
  13. Benchmarks
  14. Results
  15. Pattern A (Figure \ref{['fig:scenarioA']})
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Mission area and patterns of event detection probability change.
  • Figure 2: Pattern A: opposite decisions (top) and mission time increase (bottom) vs the knowledgeable method.
  • Figure 3: Pattern B: opposite decisions (top) and mission time increase (bottom) vs the knowledgeable method.