Michele Antonazzi

Cloud robotics allows low-power robots to perform computationally intensive inference tasks by offloading them to the cloud, raising privacy concerns when transmitting sensitive images. Although end-to-end encryption secures data in transit, it doesn’t prevent misuse by inquisitive third-party services since data must be decrypted for processing. This paper tackles these privacy issues in cloud-based object detection tasks for service robots. We propose a co-trained encoder-decoder architecture that retains only task-specific features while obfuscating sensitive information, utilizing a novel weak loss mechanism with proposal selection for privacy preservation. A theoretical analysis of the problem is provided, along with an evaluation of the trade-off between detection accuracy and privacy preservation through extensive experiments on public datasets and a real robot.

Mobile service robots are increasingly prevalent in human-centric, real-world domains, operating autonomously in unconstrained indoor environments. In such a context, robotic vision plays a central role in enabling service robots to perceive high-level environmental features from visual observations. Despite the data-driven approaches based on deep learning push the boundaries of vision systems, applying these techniques to real-world robotic scenarios presents unique methodological challenges. Traditional models fail to represent the challenging perception constraints typical of service robots and must be adapted for the specific environment where robots finally operate. We propose a method leveraging photorealistic simulations that balances data quality and acquisition costs for synthesizing visual datasets from the robot perspective used to train deep architectures. Then, we show the benefits in qualifying a general detector for the target domain in which the robot is deployed, showing also the trade-off between the effort for obtaining new examples from such a setting and the performance gain. In our extensive experimental campaign, we focus on the door detection task (namely recognizing the presence and the traversability of doorways) that, in dynamic settings, is useful to infer the topology of the map. Our findings are validated in a real-world robot deployment, comparing prominent deep-learning models and demonstrating the effectiveness of our approach in practical settings.

The multi-robot rendezvous problem requires coordinating a team of mobile robots to converge at a common location. Efficient decentralized execution with limited communication and in unknown environments can be essential in several applications, but such real-world robotic features are often not captured by theoretical rendezvous frameworks. In this work, we present an approach to this problem that extends traditional frontier-based exploration strategies to facilitate efficient rendezvous in such conditions. Our method allows robots to backtrack their exploration of the environment and exploit semantic knowledge on the map by prioritizing high-connectivity areas like corridors and hallways. We define and evaluate different variants of our method to study a trade-off between the time taken to perform a rendezvous and the amount of discovered area in the environment. Extensive experimental evaluation in ROS using 3D simulations demonstrates the feasibility of our method and its performance improvements over baselines.

We introduce a novel approach for scalable domain adaptation in cloud robotics scenarios where robots rely on third-party AI inference services powered by large pre-trained deep neural networks. Our method is based on a downstream proposal-refinement stage running locally on the robots, exploiting a new lightweight DNN architecture, R2SNet. This architecture aims to mitigate performance degradation from domain shifts by adapting the object detection process to the target environment, focusing on relabeling, rescoring, and suppression of bounding-box proposals. Our method allows for local execution on robots, addressing the scalability challenges of domain adaptation without incurring significant computational costs. Real-world results on mobile service robots performing door detection show the effectiveness of the proposed method in achieving scalable domain adaptation.

Exploration is a task in which autonomous mobile robots incrementally discover features of interest in initially unknown environments. We consider the problem of exploration for map building, in which a robot explores an indoor environment in order to build a metric map. Most of the current exploration strategies used to select the next best locations to visit ignore prior knowledge about the environments to explore that, in some practical cases, could be available. In this paper, we present an exploration strategy that evaluates the amount of new areas that can be perceived from a location according to a priori knowledge about the structure of the indoor environment being explored, like the floor plan or the contour of external walls. Although this knowledge can be incomplete and inaccurate (e.g., a floor plan typically does not represent furniture and objects and consequently may not fully mirror the structure of the real environment), we experimentally show, both in simulation and with real robots, that employing prior knowledge improves the exploration performance in a wide range of settings.