Vehicle Configuration
Customer
Our customer is one of the world's leading truck manufacturers, who designs and produces vehicles within a worldwide network. The product range covers light, medium, and heavy trucks for local and long-distance deliveries and construction sites as well as other complex special vehicles.
Challenge
Each truck is characterized by thousands of attributes, such as engine, gear-box, wheels, cabin, axles, etc. The trucks are custom made; they must be shipped according to user specification. The truck configuration must satisfy complex rules that arise from various reasons, including manufacturing, inventory, legal, pricing, policies, etc. The challenge lies in finding a valid truck configuration that meets the needs of the specific customer.
Solution
Our solution upgraded the manufacturer's configuration rules system using Stocs, the stochastic constraint solving engine developed at IBM Haifa Research Lab.
The new application is responsible for managing the complex configuration rules and the orders database. Since the rules are changing constantly, it is necessary to check that vehicles whose orders that have not yet been manufactured are still valid. The new system checks the orders periodically and finds goods alternatives for orders that are no longer valid.
The constraint solver is a key component of the new system. The configuration parameters and rules are modeled as a constraint satisfaction problem and a generic constraint satisfaction engine is used to find valid configurations. The generality of the solution is a major advantage, since it allows for flexibility in the types of possible rules.
Key features of the new system are:
- Fast testing of order validity. One of the major strengths of a stochastic constraint solver is its ability to test quickly whether a certain assignment is a solution of the constraint problem.
- Assistance with the order configuration process. Since the customer configures the desired vehicle, the system is able to suggest possible valid configurations at any stage of the configuration process.
- Finding alternatives for invalid orders. If an order became invalid as a result of a change in the configuration rules, the system is able to find valid alternatives that are as close as possible to the original order.
Achievements
- The new system will allow easy maintenance of the configuration rules, fast validation of orders, and the repair of invalid orders.
- Client is expected to gain roughly 100 orders a day that previously could not be met - at an average value of $200,000 each.
- Configuration can be validated or changed, if needed, at point of sale, providing fast response to customer's need.
- Vehicle requirements can be configured dynamically through life cycle of order, should new constraints arise after the order is placed.
In the first stage, the system is to be used by the staff documenting and managing the configuration rules. Deployment at sales points is planned at a later stage.
The vehicle configuration project has been recognized as an IBM Research Accomplishment for 2008.
Contact: Yael Ben-Haim (yaelbh@il.ibm.com)