**Problem Types:** - Vehicle Route Planning (VRP) variants: - VRP with Time Windows (VRPTW) - Capacitated VRP (CVRP) - Pick-up and Delivery VRP (PDP-VRP) - Vehicle Routing Problem with Simultaneous Pickup and Delivery (VRPSPD) - Heterogeneous fleet VRP (HVRP) - Multi-objective Dynamic Route Network Planning (MODRNP) - models the multisource to multidestination evacuation in restricted-space scenarios, with the objectives of **minimizing the whole evacuation delay** and **maximizing the evacuation efficiency.** - Studied the problem in 3D scenarios, which can provide intuition vision for the geographic space and contribute to the evacuation plan and implementation. Based on the auxiliary graph transformation, we propose a heuristic algorithm referred to the classical problem, Minimum Weighted Set Cover. - [Ref Paper here](https://www.researchgate.net/publication/326040001_Dynamic_Route_Network_Planning_Problem_for_Emergency_Evacuation_in_Restricted-Space_Scenarios) **Potential Solutions:** - Meta-heuristics that perform random search are the most promising for example, Genetic Algorithm (GA), Ant Colony System (AS), Multi-Particle Swarm Optimization (MAPSO), and Tabu Search (TS) - [Harris Hawks Optimization (HHO) algorithm ](https://journalofbigdata.springeropen.com/articles/10.1186/s40537-022-00593-4)has shown to be performant vs simulated annealing vs artificial bee colony optimisation - [Ant Colony Optimisation](https://core.ac.uk/download/pdf/225890897.pdf) - Dynamic problems have usually been solved using re-optimization (periodic re-optimization or continuous re-optimization) or fast insertion techniques heuristics with background optimization techniques [Ref here](https://link.springer.com/chapter/10.1007/978-3-642-30671-6_2) - Population-based metaheuristics: - e.g. Ant Colony, Evolutionary Algorithms, and Particle Swarm Optimization - Trajectory-based metaheuristics - e.g. Tabu Search, Greedy Randomized Adaptive Search Procedure (GRASP), variable neighborhood search - Deep RL model: https://paperswithcode.com/paper/a-deep-reinforcement-learning-algorithm-using - Deep Policy Dynamic Programming: https://paperswithcode.com/paper/deep-policy-dynamic-programming-for-vehicle - Handling the large search space: https://paperswithcode.com/paper/dynamic-partial-removal-a-neural-network  **DYNAMIC INPUTS:** Dynamic problems where the existence of the input data is known but the e*xact time of their availability is not known in advance*. This availability may depend on the state of the system and can be determined using a general rule specifying any dependencies between input data (for example rule according to which it can be determined from when it is possible to deliver to the given company). It is impossible to determine a priori values of all variables and parameters for this class of problems. In such problems, not all decisions are possible at every stage due to the existing constraints or dynamic input data. In consequence, it is hard or impossible to simply implement the improvement algorithms because a significant number of obtained solutions will not be allowed. Thus, such problems require a distinct approach that allows one to construct the solution sequentially and significantly reduce the number of impossible solutions. The **decision-making process should be carried out in stages**. An approach that meets these needs is the Algebraic-Logical Meta-Model (ALMM). The ALMM approach is based on a multistage decision-making process, where the decision is made jointly and considers the current situation (current process state). It is a trajectory-based simulation. Previously, the ALMM approach has been used for a dynamic vehicle routing problem [[22](https://www.mdpi.com/2073-8994/11/4/546#B22-symmetry-11-00546)] and the supply routes for a multi-location companies problem [[23](https://www.mdpi.com/2073-8994/11/4/546#B23-symmetry-11-00546)]. This paper presents an approach based on the ALMM methodology for a dynamic vehicle routing problem with the possibility of modeling the dynamic predictive events, in particular, the appearance of new customers to visit during the simulation according to an appropriate general rule. [Ref here](https://www.mdpi.com/2073-8994/11/4/546) Interesting Read on Bus Operation strategy definition in a evacuation context: https://onlinelibrary.wiley.com/doi/epdf/10.1002/atr.1224 and a dissertation on the Bus Evacuation Problem https://vtechworks.lib.vt.edu/bitstream/handle/10919/52361/Pereira_VC_D_2013.pdf;sequence=1 See data fusion method here and also dispatching and routing simulation model: http://www.acsu.buffalo.edu/~batta/papers/Jotshi_Gong_Batta_SEPS.pdf Deep Dive into VRP Solutions: https://bib.irb.hr/datoteka/433524.Vehnicle_Routing_Problem.pdf **Comparison Criteria:** - minimum objective function obtained - minimum number of iterations required - satisfaction of capacity constraints **Rough References** See how the penalty severity could evolve over time based on the nature of the emergency inducing event. ![[Pasted image 20221212234326.png]] See [Ref here](https://www.sciencedirect.com/science/article/pii/S2212420920312826) ![[Pasted image 20221213000835.png]] ![[Pasted image 20221213000856.png]] ![[Pasted image 20221213001044.png]] ![[Pasted image 20221213001103.png]] ##### Evacuation Sequence The sequence of an evacuation can be divided into the following phases: 1. detection 2. decision 3. alarm 4. reaction 5. movement to an area of refuge or an assembly station 6. transportation