Motion Planning with Temporal Goals
Traditional motion planning algorithms have considered
the problem of constructing a motion plan for a given robot
model, such that the plan takes the robot from a set of initial
states to a set of goal states while avoiding obstacles in the
workspace. A class of planning approaches have been
proposed recently that use a richer framework to specify
complex temporal goals (for example, coverage or ordering of events) using formalisms such as Linear Temporal Logic
(LTL). The following is an example of a complex temporal goal:
In the future, visit region p1, and then visit region p2 or region p3.
Planning Framework:
We use a multi-layered synergistic framework that has been proposed independently for safety analysis of hybrid systems with LTL specifications (TemporalHyDICE). The multi-layered synergistic planning paradigm is a very general and powerful paradigm that has also shown to be remarkably efficient for solving traditional motion planning (SyCLoP) and safety analysis (HyDICE) problems. Our instantiation of the framework is shown above and consists of three main layers:
a) High-level search layer: This is the layer that uses a discrete abstraction for the robot model, the specifications, and the exploration information from the low-level search layer to construct high-level plans. In contrast to existing approaches, this layer differs in the fact that it incorporates the exploration information from the low-level search layer as well.
b) Low-level search layer: The low-level search layer is the layer that accounts for the physics of the problem. This layer takes into account the robot dynamics and the geometric constraints while exploring the physical space for feasible trajectories. The layer uses suggested high-level plans to explore the state- space for solution trajectories.
c) Synergy layer: The synergy layer is an intermediate layer that facilitates the synergistic interaction between the high-level and the low-level search layers. This layer is a critical component of the overall framework.
Current Focus:
Previous work focused on exploring the role of synergy and did not consider the issue of construction of the discrete abstraction. In fact, the discrete abstraction was assumed to be a user-supplied input within the framework. The discrete abstraction plays a critical role in the effectiveness and performance of the overall approach. We are currently investigating the issue of construction and use of the discrete abstraction in the high-level search layer. So far we have proposed geometry-based techniques for the construction of the discrete abstraction. We have also proposed a lazy search technique for high-level exploration.
For the construction of the discrete abstraction, we have considered two cases:
a). Case of continuous robot dynamics
b). Case of hybrid robot dynamics
In both cases we have proposed geometry-based decompositions that use a triangulation of the workspace. Our proposed approach shows performance improvement of up to 10 times.
This work has been supported by CNS 0615328.
Planning Framework:
We use a multi-layered synergistic framework that has been proposed independently for safety analysis of hybrid systems with LTL specifications (TemporalHyDICE). The multi-layered synergistic planning paradigm is a very general and powerful paradigm that has also shown to be remarkably efficient for solving traditional motion planning (SyCLoP) and safety analysis (HyDICE) problems. Our instantiation of the framework is shown above and consists of three main layers:
a) High-level search layer: This is the layer that uses a discrete abstraction for the robot model, the specifications, and the exploration information from the low-level search layer to construct high-level plans. In contrast to existing approaches, this layer differs in the fact that it incorporates the exploration information from the low-level search layer as well.
b) Low-level search layer: The low-level search layer is the layer that accounts for the physics of the problem. This layer takes into account the robot dynamics and the geometric constraints while exploring the physical space for feasible trajectories. The layer uses suggested high-level plans to explore the state- space for solution trajectories.
c) Synergy layer: The synergy layer is an intermediate layer that facilitates the synergistic interaction between the high-level and the low-level search layers. This layer is a critical component of the overall framework.
Current Focus:
Previous work focused on exploring the role of synergy and did not consider the issue of construction of the discrete abstraction. In fact, the discrete abstraction was assumed to be a user-supplied input within the framework. The discrete abstraction plays a critical role in the effectiveness and performance of the overall approach. We are currently investigating the issue of construction and use of the discrete abstraction in the high-level search layer. So far we have proposed geometry-based techniques for the construction of the discrete abstraction. We have also proposed a lazy search technique for high-level exploration.
For the construction of the discrete abstraction, we have considered two cases:
a). Case of continuous robot dynamics
b). Case of hybrid robot dynamics
In both cases we have proposed geometry-based decompositions that use a triangulation of the workspace. Our proposed approach shows performance improvement of up to 10 times.
This work has been supported by CNS 0615328.