# SEArch: Optimistic Policy Selection Between Scene Noise and Drift for UAV Radar Search

arXiv: https://arxiv.org/abs/2606.01325 · HTML: https://mlsanity.com/paper/2606.01325/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 45 references that are in the map.
- First posted on arXiv 31 May 2026.

## Builds on (most cited first, 19 of 19)

- [Introduction to Online Convex Optimization](https://mlsanity.com/paper/1909.05207/) (2019)
- [Online Learning: A Modern Introduction Using Convex Optimization](https://mlsanity.com/paper/1912.13213/) (2019)
- [Online Learning: A Comprehensive Survey](https://mlsanity.com/paper/1802.02871/) (2018)
- [Online Meta-Learning](https://mlsanity.com/paper/1902.08438/) (2019)
- [Adaptive Online Learning in Dynamic Environments](https://mlsanity.com/paper/1810.10815/) (2018)
- [Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments](https://mlsanity.com/paper/1710.03641/) (2017)
- [Online Optimization : Competing with Dynamic Comparators](https://mlsanity.com/paper/1501.06225/) (2015)
- [A Second-order Bound with Excess Losses](https://mlsanity.com/paper/1402.2044/) (2014)
- [Machine Learning-Aided Operations and Communications of Unmanned Aerial Vehicles: A Contemporary Survey](https://mlsanity.com/paper/2211.04324/) (2022)
- [Understanding Adam Optimizer via Online Learning of Updates: Adam is FTRL in Disguise](https://mlsanity.com/paper/2402.01567/) (2024)
- [Between Stochastic and Adversarial Online Convex Optimization: Improved Regret Bounds via Smoothness](https://mlsanity.com/paper/2202.07554/) (2022)
- [Online Caching with Optimistic Learning](https://mlsanity.com/paper/2202.10590/) (2022)
- [Bandit Submodular Maximization for Multi-Robot Coordination in Unpredictable and Partially Observable Environments](https://mlsanity.com/paper/2305.12795/) (2023)
- [On the Dynamic Regret of Following the Regularized Leader: Optimism with History Pruning](https://mlsanity.com/paper/2505.22899/) (2025)
- [Parameter-free Algorithms for the Stochastically Extended Adversarial Model](https://mlsanity.com/paper/2510.04685/) (2025)
- [Reinforcement Learning for Agile Active Target Sensing with a UAV](https://mlsanity.com/paper/2212.08214/) (2022)
- [Partially Lazy Gradient Descent for Smoothed Online Learning](https://mlsanity.com/paper/2601.15984/) (2026)
- [Online Trajectory Optimization Using Inexact Gradient Feedback for Time-Varying Environments](https://mlsanity.com/paper/2001.00685/) (2020)
- [PDSR: Efficient UAV Deployment for Swift and Accurate Post-Disaster Search and Rescue](https://mlsanity.com/paper/2410.22982/) (2024)

## Abstract (arXiv)

Unmanned Aerial Vehicles (UAVs) equipped with radar sensors are deployed for target search missions in diverse environments, where targets exhibit characteristic signatures (e.g., respiration micro-motion in human search) detectable through occlusions. A fundamental challenge arises from shifts in radar statistics as the UAV moves through a dynamic and potentially non-stationary environment, rendering any fixed signal-processing strategy suboptimal; yet perception and adaptation must run onboard a resource-constrained aerial node in real time. Since no single detector performs well across all conditions, we adopt a multi-policy paradigm and formulate UAV target search as an online policy selection problem over a library of specialized detectors, with performance measured by regret, the cumulative loss gap relative to the best policy in each scene. The setting couples in-scene stochastic noise with inter-scene shifts. Whereas prior methods capture only one regime, we account for both through the Stochastically Extended Adversary (SEA) framework, without requiring oracle knowledge of scene dynamics. Because adaptation must run at the UAV, we instantiate SEA through textscSEArch, a lightweight optimistic Follow the Regularized Leader (OFTRL) selector with an adaptive learning rate, achieving regret O(barσT √T + √J), where barσT captures radar measurement noise and J is the number of scene transitions over the mission horizon T. To enable rapid adaptation under frequent scene changes, we further introduce textscW-SEArch, a windowed variant that restarts every w rounds and achieves regret O(barσI √w) under at most one transition per window. Experiments show up to 30% regret reduction compared to non-adaptive baselines across a range of non-stationary settings.