Similar appearing places are known to confuse place recognition methods (a.k.a perceptual aliasing). Robust visual SLAM methods have shown impressive resilience against scenarios having moderate perceptual aliasing. In this work, we evaluate visual SLAM in scenarios where perceptual aliasing occurs with greater frequency. Firstly we evaluate visual SLAM front-ends. We modify the environment by replicating a simple high textured patch. Surprisingly, this simple patch manages to by-pass all the geometric checks in the state of the art ORB-SLAM, resulting in false loop closures and hence corrupting the map. The patch can be printed on a simple paper, using a simple printer, or it can be digital content displayed on screens. We show that closeups (approaching doors, turning a corridor, interacting with an object) are the most vulnerable situations. This vulnerability exists in multiple loop closing pipelines. Secondly, we evaluate robust SLAM back-ends. These back-ends have shown the ability to recover from false positives under multiple policies (random, local, random grouped and locally grouped). We propose a novel policy, i.e. locally symmetric, for generating false loop closures which successfully attacks multiple robust back-ends and public datasets. Surprisingly, for commonly occurring weaker structures (multiple floors) even a single false loop manages to fool multiple robust back-ends. We hope these findings will motivate the community in evolving SLAM solutions for adversarial environments. To further enhance progress in this direction, we release a novel dataset with adversarial content in both visual SLAM front-ends and back-ends 1 .
This paper presents a sensor-data-fusion method using an Unscented Kalman Filter (UKF), to implement an accurate localization system for mobile robots. Integration of data from various sensors using an efficient sensor fusion algorithm is required to achieve a continuous and accurate localization of mobile robots. We use data from low cost accelerometer, gyroscope, and encoders to obtain robot motion information. The UKF, used as an efficient sensor fusion algorithm, is an advanced filtering technique which outperforms the widely-used Extended Kalman Filter (EKF) in many applications. The system is able to compensate for the slip errors by switching between two different UKF models built for slip and no-slip cases. Since the accelerometer error accumulates with time because of the double integration, the data from accelerometer is only used in slip model of the UKF. The results obtained from UKF sensor fusion algorithm are compared with the results from an accurate distance laser sensor. The experimental results show that the system is able to accurately track the motion of the robot in various robot motion scenarios including the scenario when robot's encoders data is not reliable due to the slip occurrence.