Robust Execution Strategies for Probabilistic Temporal Planning

Consider a scheduling scenario in a furniture factory where two robotic agents work together to apply finish to custom table pieces. Robot A is set up to paint, while Robot B must apply varnish. Both actions take an uncertain amount of time due to physical variation in the size and shape of the table components. Painting takes around six minutes to complete, while applying varnish take around two minutes. To maintain efficiency at the factory, the tasks must be completed within two minutes of each other before the furniture components must move on. This scheduling problem is illustrated in Figure 1. Successfully executing this task requires coordination between the robots—if both robots started at the same time, Robot B is likely to finish varnishing its component before Robot A finishes painting. The challenge that this abstract addresses is determining the best way to schedule events in the face of scheduling uncertainty.