We present the design, implementation and information flow verification of CoSMeDis, a distributed social media platform. The system consists of an arbitrary number of communicating nodes, deployable at different locations over the Internet. Its registered users can post content and establish intra-node and inter-node friendships, used to regulate access control over the posts. The system's kernel has been verified in the proof assistant Isabelle/HOL and automatically extracted as Scala code. We formalized a framework for composing a class of information flow security guarantees in a distributed system, applicable to input/output automata. We instantiated this framework to confidentiality properties for CoSMeDis's sources of information: posts, friendship requests, and friendship status.

In this paper we present an affordable solution to help blind people navigate unknown environments. Our solution performs image classification on a Raspberry Pi and provides feedback to users by means of vibration motors to signal the presence of an obstacle in a given direction. The training phase is performed off-line, while the on-line phase can classify an image in 1.12 seconds on average. We provide an evaluation using several thousands images, showing that we can achieve a precision of 79% and a recall of 79%. All our code and the hardware design files are released open source.

No abstract available.

Small-footprint mobile ground robots, such as the popular Turtlebot and Kobuki platforms, are by necessity equipped with sensors which lie close to the ground. Reliably detecting and tracking people from this viewpoint is a challenging problem, whose solution is a key requirement for many applications involving sharing of common spaces and close human-robot interaction. We present a robust solution for cluttered indoor environments, using an inexpensive RGB-D sensor such as the Microsoft Kinect or Asus Xtion. Even in challenging scenarios with multiple people in view at once and occluding each other, our system solves the person detection problem significantly better than alternative approaches, reaching a precision, recall and F1-score of 0.85, 0.81 and 0.83, respectively. Evaluation datasets, a real-time ROS-enabled implementation and demonstration videos are provided as supplementary material.