Tiny Robotics Dataset and Benchmark for Continual Object Detection

2024-09-24

arXiv.org

AI Summary

To address the challenge of achieving robust, continual object detection on micro-mobile robots under severe constraints in size, power consumption, and computational capacity, this paper introduces the first lightweight continual learning benchmark for object detection. We propose TiROD—a novel dataset collected directly from real robotic platforms—and establish the first evaluation paradigm jointly optimizing computational efficiency, energy consumption, and generalization capability for lightweight continual detection. Building upon NanoDet as the baseline architecture, we systematically integrate replay, regularization, and architectural expansion strategies, and conduct rigorous incremental class- and domain-adaptation evaluation on realistic visual streams. Experiments reveal two critical bottlenecks: severe catastrophic forgetting and poor cross-domain transferability in lightweight detectors. Our work provides a reproducible benchmark, a standardized evaluation protocol, and strong baseline results—laying foundational groundwork for continual intelligence evolution in resource-constrained edge systems.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Machine Learning: Life-Long and Continual Learning
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Detecting objects in mobile robotics is crucial for numerous applications, from autonomous navigation to inspection. However, robots often need to operate in different domains from those they were trained in, requiring them to adjust to these changes. Tiny mobile robots, subject to size, power, and computational constraints, encounter even more difficulties in running and adapting these algorithms. Such adaptability, though, is crucial for real-world deployment, where robots must operate effectively in dynamic and unpredictable settings. In this work, we introduce a novel benchmark to evaluate the continual learning capabilities of object detection systems in tiny robotic platforms. Our contributions include: (i) Tiny Robotics Object Detection~(TiROD), a comprehensive dataset collected using the onboard camera of a small mobile robot, designed to test object detectors across various domains and classes; (ii) a benchmark of different continual learning strategies on this dataset using NanoDet, a lightweight object detector. Our results highlight key challenges in developing robust and efficient continual learning strategies for object detectors in tiny robotics.

Problem

• Evaluating continual learning in tiny robotics object detection
• Adapting object detection to dynamic, unpredictable environments
• Developing efficient algorithms for resource-constrained tiny robots

Innovation

• Tiny Robotics Object Detection (TiROD) dataset
• Benchmark for continual learning strategies
• Lightweight object detector NanoDet used