We present a framework for combining imitation learning and logical automata and introduce the Logic-based Value Iteration Network (LVIN) model. By appending a ‘logical’ dimension to the state space of the environment, LVIN can recover and incorporate the transition matrix of a finite state automaton derived from a Linear Temporal Logic formula into the policy learned through imitation. This approach improves upon the original VIN in several ways: LVIN (1) is capable of learning logic corresponding to long sequences of steps, (2) adapts easily to new specifications, and (3) is amenable to correction after faulty expert demonstrations (e.g., in a driving domain).
This work focuses on integrated routing and motion planning for an autonomous vehicle in a road network. We consider a problem in which customer demands need to be met within desired deadlines, and the rules of the road need to be satisfied. The vehicle might not, however, be able to satisfy these two goals at the same time. We propose a systematic way to compromise between delaying the satisfaction of the given demand and violating the road rules. We utilize scLTL formulas to specify desired behavior and develop a receding horizon approach including a periodically interacting routing algorithm and a RRT*-based motion planner. The proposed solution yields a provably minimum-violation trajectory. An illustrative case study is included.
Synthetic biology is founded on the central dogma of molecular biology, which describes flow of genetic information from DNA to RNA to Protein. Using this foundation as a platform for logic synthesis, biologists have engineered genetic modules that can resemble logical functions such as a NOR gate (logical joint denial). Although a NOR gate is a universal gate (it can be used to implement any combinatorional logic circuit without the need to use any other gate type), Boolean functions lack the functional richness required to capture the finer details of molecular biology. Furthermore, sparse or poorly defined characterization of genetic modules makes creating robust genetic circuits difficult. To solve these problems, this work uses the principles of systems engineering, and introduces a framework to build reliable and robust genetic systems.
We developed a sampling-based motion planning algorithm that combines long-term temporal logic goals with short-term reactive requirements. The mission specification has two parts: (1) a global specification given as a Linear Temporal Logic (LTL) formula over a set of static service requests that occur at the regions of a known environment, and (2) a local specification that requires servicing a set of dynamic requests that can be sensed locally during the execution. The proposed computational framework consists of two main ingredients: (a) an off-line sampling-based algorithm for the construction of a global transition system that contains a path satisfying the LTL formula, and (b) an on-line sampling-based algorithm to generate paths that service the local requests, while making sure that the satisfaction of the global specification is not affected. The off-line algorithm has three main features. First, it is incremental, in the sense that the procedure for finding a satisfying path at each iteration scales only with the number of new samples generated at that iteration. Second, the underlying graph is sparse, which guarantees the low complexity of the overall method. Third, it is probabilistically complete. We also provide a conditional result showing that the incremental checking procedure has the best possible complexity bound. The on-line algorithm leverages ideas of potential functions, which ensure progress towards satisfaction of the global specification, and on monitors for LTL. Examples illustrating the usefulness and the performance of the framework are included.