Abstract This presentation shows an automatic task and motion planning framework for multi-robot systems (MRS) to achieve a temporal logic constrained task by considering each robot’s trustworthiness based on the automaton theory. Given a set of temporal logic formulae described task specifications for an MRS, we first develop an automaton based iterative parallel decomposition framework thus obtain the largest set of parallel performable subtasks. These automata described subtasks are then assigned into each robot inside the MRS team by considering the costs (or rewards) of completing a single task as well as the overall concurrency. The costs (or rewards) here are quantified by the trust value of the corresponding robots, which is a probabilistic estimation on a robot by considering multi-dimensional metrics, such as performance of a robot and supervision workload from an operator. A dynamic Bayesian network (DBN) is utilized to catch the evolution of trust. Besides, robots are configured with partially overlapping capabilities so that the ones with the appropriate trust value will be selected. The final robot-task assignment plan, i.e., a task plan, satisfies both the temporal logic constraints and optimality of robot configuration. Symbolic motion planning (SMP) is performed for each individual robot based on the task plan in a discretized environment. Task redecomposition and replanning are triggered in a distributed manner to update the task plan and SMP among neighboring robots.