An Improved Genetic Algorithm for Vehicle Routing Problem with Hard Time Windows

When the logistic industry plans for delivering the goods to the customers, travel cost reduction is one aspect that is crucial to consider. Among the popular Vehicle Routing Problem (VRP) variants addressing the travel cost minimization, Vehicle Routing Problem with Time Windows (VRPTW) is one of the most fundamental and practical variants. This study proposes a new improved Genetic Algorithm (GA) to solve the hard time windows variant of VRPTW by developing the problem-specific crossover and seven different mutation operators. One of the mutations uses the heuristics information to guide the GA when it explores the new features in the large solution space. This design contributes efficiently to the randomly generated customer coordinates. The performance of the proposed GA is assessed on the well-known Solomon benchmarks which contain 100 customers for each instance. The results from our GA are improved as it is competitive and better than the best-known solutions from the previous studies.