The automotive industry, a cornerstone of modern economies, faces mounting challenges in optimizing its production systems amidst growing complexity, sustainability demands, and just-in-time (JIT) delivery pressures. This dissertation focuses on addressing key optimization problems within automotive manufacturing, particularly in assembly line feeding and tow train routing, to enhance efficiency and reduce operational costs. The primary focus of the dissertation is the assembly line feeding problem (ALFP), a critical component of JIT systems where tow trains deliver parts to workstations. Traditional optimization methods often struggle to solve large-scale ALFP instances effectively. To address this, a variable neighborhood search (VNS) algorithm is proposed, incorporating the split delivery vehicle routing problem (SDVRP) framework. This approach minimizes travel distances and reduces the number of required tow trains while analyzing the interplay between cycle time, delivery volumes, and split delivery costs in a multi-objective problem setting. The results provide actionable strategies for optimizing real-world ALFP systems. The second focus tackles conflict-free tow train routing, a persistent challenge in JIT manufacturing systems. Conflicts among tow trains affect both safety and operational efficiency. To address this, the production layout is partitioned into a grid of “pixels,” enabling the detection of potential collisions. Shortest paths between workstations are computed using the A-star (A*) algorithm, and a simulated annealing (SA) heuristic optimizes tow train routes to meet workstation demands. A conflict detection algorithm (CDA) identifies and resolves collisions through two distinct avoidance strategies, ensuring safe and efficient routing. Benchmark tests demonstrate the effectiveness of this approach. The final focus addresses tow train routing in narrow-aisle environments, where vehicles cannot reverse in place and risk blocking one another.\n\nThe workspace is represented as a pixel-based grid, allowing for the precise detection of blocking vehicles. The shortest paths between workstations are determined using Dijkstra’s algorithm, and the routing problem is formulated as a generalized vehicle routing problem (GVRP). A mathematical model is developed, and a simulated annealing (SA)-based heuristic with problem-specific neighborhood operators is implemented. A blocking detection and avoidance algorithm resolves vehicle conflicts. Computational studies confirm the heuristic’s ability to find efficient solutions in these challenging scenarios. By combining advanced optimization techniques, conflict management strategies, and meta-heuristic frameworks, this dissertation significantly advances operational efficiency in automotive manufacturing. The findings provide scalable and practical solutions to industry-relevant challenges, contributing to the state of the art in JIT manufacturing and vehicle routing optimization.