Ana Sofia Rufino Ferreira

Ana Sofia Rufino Ferreira, Justin Hsia, Murat Arcak et al.

We propose a compartmental lateral inhibition system that generates contrasting patterns of gene expression between neighboring compartments. The system consists of a set of compartments interconnected by channels. Each compartment contains a colony of cells that produce diffusible molecules to be detected by the neighboring colony, and each cell is equipped with an inhibitory circuit that reduces its production when the detected signal is stronger. We develop a technique to analyze the steady-state patterns emerging from this lateral inhibition system and apply it to a specific implementation. The analysis shows that the proposed system indeed exhibits contrasting patterns within realistic parameter ranges.

Ana S. Rufino Ferreira, Murat Arcak

We analyze pattern formation on a network of cells where each cell inhibits its neighbors through cell-to-cell contact signaling. The network is modeled as an interconnection of identical dynamical subsystems each of which represents the signaling reactions in a cell. We search for steady state patterns by partitioning the graph vertices into disjoint classes, where the cells in the same class have the same final fate. To prove the existence of steady states with this structure, we use results from monotone systems theory. Finally, we analyze the stability of these patterns with a block decomposition based on the graph partition.

Stefano Pini, Christian S. Perone, Aayush Ahuja et al.

The goal of autonomous vehicles is to navigate public roads safely and comfortably. To enforce safety, traditional planning approaches rely on handcrafted rules to generate trajectories. Machine learning-based systems, on the other hand, scale with data and are able to learn more complex behaviors. However, they often ignore that agents and self-driving vehicle trajectory distributions can be leveraged to improve safety. In this paper, we propose modeling a distribution over multiple future trajectories for both the self-driving vehicle and other road agents, using a unified neural network architecture for prediction and planning. During inference, we select the planning trajectory that minimizes a cost taking into account safety and the predicted probabilities. Our approach does not depend on any rule-based planners for trajectory generation or optimization, improves with more training data and is simple to implement. We extensively evaluate our method through a realistic simulator and show that the predicted trajectory distribution corresponds to different driving profiles. We also successfully deploy it on a self-driving vehicle on urban public roads, confirming that it drives safely without compromising comfort. The code for training and testing our model on a public prediction dataset and the video of the road test are available at https://woven.mobi/safepathnet