Research

Twist-n-Sync: Software Clock Synchronization with Microseconds Accuracy Using MEMS-Gyroscopes

Sensor networks require a high degree of synchronization in order to produce a stream of data useful for further purposes. Examples of time misalignment manifest as undesired artifacts when doing multi-camera bundle-adjustment or global positioning system (GPS) geo-localization for mapping. Network Time Protocol (NTP) variants of clock synchronization can provide accurate results, though present high variance conditioned by the environment and the channel load. We propose a new precise technique for software clock synchronization over a network of rigidly attached devices using gyroscope data. Gyroscope sensors, or IMU, provide a high-rate measurements that can be processed efficiently. We use optimization tools over the correlation signal of IMU data from a network of gyroscope sensors. Our method provides stable microseconds accuracy, regardless of the number of sensors and the conditions of the network. In this paper, we show the performance of the gyroscope software synchronization in a controlled environment, and we evaluate the performance in a sensor network of smartphones by our open-source Android App, Twist-n-Sync, that is publicly available.

Precise synchronization allows to capture the same snow-flakes from two cameras.

 

Eigen-Factors

The Eigen-Factor (EF) method, which estimates a planar surface from a set of point clouds (PCs), with the peculiarity that these points have been observed from different poses, i.e. the trajectory described by a sensor. We propose to use multiple Eigen-Factors (EFs) or different planes’ estimations, that allow to solve the multi-frame alignment over a sequence of observed PCs. Moreover, the complexity of the EFs optimization is independent of the number of points, but depends on the number of planes and poses. To achieve this, a closed-form of the gradient is derived by differentiating over the minimum eigenvalue with respect to poses, hence the name Eigen-Factor.

Adversarial planning via MPDM

We proposed a new method for navigation amongst pedestrians in which the trajectory of the robot is not explicitly planned, but instead, a planning process selects one of a set of closed-loop behaviors whose utility can be predicted through forward simulation. In particular, we extend Multi-Policy Decision Making (MPDM) to this domain using the closed-loop behaviors Go-Solo, Follow and Stop.

By dynamically switching between these policies, we show that we can improve the performance of the robot as measured by utility functions that reward task completion and penalize inconvenience to other agents. In MPDM a robot’s policy is elected by sampling from the distribution of current states.

Discovering potentially dangerous configurations through random sampling may require drawing many samples, which becomes a performance bottleneck. We described a risk-aware approach which augments this sampling with an optimization process that helps discover those influential outcomes.