SHF: Medium: Automating robot programming through constraint solving and motion planning

该项目旨在为个人机器人开发一个名为RobSynth的高级编程框架。这里，机器人的用户不是编写定义机器人必须如何执行任务的低级代码，而是编写定义要完成什么的规范。在给定该规范和机器人环境模型的情况下，RobSynth自动合成可在机器人上执行的程序。只要环境按照假设的模型运行，这个程序的所有执行都可以保证满足用户定义的需求。这种方法及其衍生方法可以使大量未被开发的程序员能够访问机器人编程。该项目的技术亮点是用户与RobSynth交互所使用的规范语言，以及RobSynth用于自动代码合成的算法。这些算法同时推理出与机器人的高级目标有关的逻辑任务级别，以及与导航和操纵物理空间有关的连续运动级别。在任务级别，RobSynth利用最新的方法来分析复杂的逻辑约束系统，例如SMT求解和图论游戏的符号解。运动级推理使用基于采样的运动规划技术来执行。

The project aims to develop a high-level programming framework, called Robosynth, for personal robots. Here, rather than writing low-level code that defines how a robot must perform a task, the user of the robot writes a specification that defines what is to be accomplished. Given this specification and a model of the robot's environment, Robosynth automatically synthesizes a program that can be executed on the robot. So long as the environment behaves according to the assumed model, all executions of this program are guaranteed to satisfy the user-defined requirements.This approach and its derivatives can make robot programming accessible to a vast untapped body of inexperienced programmers.The technical highlights of the project are the specification language using which users interact with Robosynth, and the algorithms that Robosynth uses for automatic code synthesis. These algorithms simultaneously reason about a logical task level that is concerned with the high-level goals of the robot, as well as a continuous motion level concerned with navigating and manipulating a physical space. At the task level, Robosynth leverages recent methods for analyzing complex systems of logical constraints, for example SMT-solving and symbolic solution of graph games. Motion-level reasoning is performed using sampling-based motion planning techniques.