TOSCA

TOSCA

Description : The Opportunistic Job-Shop Scheduling System

Client : Hitachi Ltd

Problem: Hitachi receive a large number of requests for factory scheduling systems. This in itself makes scheduling an important problem to the company, more so given the potential cost saving involved if good scheduling solutions can be achieved. Most current commercial scheduling systems which apply an essentially simple dispatch-rule approach have various limitations most significantly in terms of their ability to satisfy multiple, potentially conflicting objectives, such as meeting order due dates and restricting the overall work in process.

Solution: TOSCA addresses the factory scheduling problem from a constraint-based perspective. Constraints are monitored in terms of potential threats to their satisfaction and are opportunistically tackled through a process which iteratively refines the schedule by restricting the resourcing and start time options.

Main Features: TOSCA provides

a pre-schedule analysis which serves to provide an understanding of the feasibility of the problem in terms of its constraints
a high-level schedule which allocates factory operations to a single resource and a restricted time period taking preferences into account. This output can be treated as a full schedule which leaves responsibility for detailed scheduling (i.e. choosing the specific start time for operations) under human control.
a detailed schedule which allocates factory operations to a single resource and specific start time.

Technical Details: TOSCA has been implemented in Common Lisp and CLOS.

Development Effort: The total combined effort for the project is 5 years over a period of 2.5 years.

TOSCA papers available

Constraint monitoring in TOSCA (AIAI Technical Report No 118)
AAAI-92 workshop paper
The Management of Job-Shop Scheduling Constraints (AIAI Technical Report No 121)
NSF Dynamic Scheduling Workshop paper, Jan 1993
Domain Description Language (AIAI Technical Report No 137, Oct 1993)

Return to AIAI information