ColeridgeGuo / Summer-Research-2018

A* Search Algorithm for the Improved Spatial Navigation System for the Visually Impaired Students

Summer 2018 Project

InsertCSV

• First off, import *_Edges.csv, *_Nodes.csv, and *_Connections.csv into tEdgeAttributes, tNodeAttributes, and tNode_Edge_Connections tables, respectively (skip the first row of the files, which are headers).
  • All indices must start at 0.
  • *_Edges.csv contains all the edges and their attributes.
    • *_Edges.csv has to have the columns in the order of: edgeID, Location, PathType, StairNum, Curb, Traffic, Obstacles, ExtraCost. Names of the columns are not required to be exactly the same as above. edgeID and StairNum are both of type int; ExtraCost is of type double; the rest of the columns are of type varchar(100).
  • *_Nodes.csv contains all the nodes and their attributes.
    • *_Nodes.csv has to have the columns in the order of: nodeId, I_E, Latitude, Longitude, Altitude, NodeType, Location, RoomNumber, Floor, Description, DoorLocation, Swipe, OpenDirection. Names of the columns are not required to be exactly the same as above. nodeID and Floor are both of type int; Latitude, Longitude, and Altitude are of type double; I_E is of type char; the rest of the columns are of type varchar(100).
  • *_Connections.csv contains all the connections between nodes.
    • *_Connections.csv has to have the columns in the order of: nodeID, Location, edgeID, Latitude, Longitude, and Altitude. Names of the columns are not required to be exactly the same as above. nodeID and edgeID are both of type int; Latitude, Longitude, and Altitude are of type double; Location is of type varchar(100).
• Then run importCSV.php. It finds all the matches between nodes and their based on the connections. It skips any nodes that have been already inserted in the database so that no duplicates exist. The final Nodes and Edges tables contain all nodes and edges without duplicates, and NodeAttributes, and EdgeAttributes table contain any information about them. It then clears out all the "t" tables.
• TODO: Add connections between layers. Edges between the insides and the outsides of doors and edges that connect stairs and elevators have to be added into the database (currently manually added).
• Run updateEdgeCost.php after all the edges, including the manually added edges, are inserted into the database to calculate their lengths and extra costs.
• Thus we have established a database of all nodes and edges for path-finding.

AStar_php

astar_node.php

• Privately used by aStarPath.php.

• A Node object has attributes such as id, latitude, longitude, which are self-explanatory. The following attributes are used in the A* search algorithm:g is the current cost of the node; h is a heuristic (the Euclidean distance between the node and the target in this implementation); f is the current cost plus the heuristic; parent is the node from which the path comes to the current node; adjacencies is an array of all the edges connected to the current node.

• The method distance2segment projects the current node onto the edge bounded by endPointA and endPointB, finds the distance from the node to the edge, which is a line segment. It returns the distance between the node and one of the endpoints if the projection of the node lies outside of the edge. It also returns the closest point on edge to the node if required.

astar_edge.php

• Privately used by aStarPath.php.

• An Edge object resembles an "edge" in the database. It has two attributes of type Node, endPointA and endPointB, along with the cost of the edge.

• The method getOther returns the other endpoint that is not the one passed as the parameter.

aStarPath.php

• The main path-finding algorithm, using the A* shortest-path algorithm as the underlying structure.
• The function createStart gets the latitude and longitude of the user's current location from HTTP POST requests, creates a starting node, and stores it in the database. It then calls the nearestEdge method which returns the minimum distance between the current position and its nearest edge as well as the closest edge itself. The distance2segment method of the Node class get called to find the closest point on the edge. If the closest point is either endpoint of the edge, only the temporary edge in between the current position and the endpoint is added into the database; if the closest point lies inside the edge, then it gets added into the database along with three temporary edges: the edge between the current position and the closest point on edge, and the two edges between the closest point and the two endpoints of the edge.
• The function nearestEdge looks at the top five closest nodes of the parameter node, and looks at all edges of each node to find the nearest edge and distance between the parameter node and the edge.
  • If the current location is inside, it only looks for the top five interior nodes to find the closest node on edge; if it is outside, it looks at the top five exterior nodes.
• The function createTarget gets the building name and room number from HTTP POST requests, and creates a destination node.
• The function AStarSearch is the main implementation of the A* path-finding algorithm.
• The function buildAdjacencies builds the adjacency array of the node for every edge that the node is attached to.
• The function printPath follows the path from destination, via the parent of each node, back to the start, and reverses it as the path. A JSON string with all the information of each node and edge on the path is returned for the APP's use.
• The function clearTempNodesNEdges clears the temporary edges and nodes for the current user.

MySQL Database

Nodes

• The Nodes table contains all the nodes in the map.

• The Nodes table contains nodeID, I_E, Temporary, Latitude, Longitude, Altitude, MagX, MagY, MagZ. The primary key is nodeID. I_E stands for "whether the node is an interior or exterior node" and is of type char. Temporary is an attribute for a temporary node created during the initialization of the search algorithm. MagX, MagY, and MagZ are three parameters for the magnetic measurements (To be used).

Edges

• The Edges table contains all the edges in the map.
  • For exit-only edges, PathType has to be "Exit only", and Obstacles has to be either "A to B" or "B to A" depending on the direction of the one-way path.

NodeAttributes

• The NodeAttributes table contains all the attributes of each node.

EdgeAttributes

• The EdgeAttributes table contains all the attributes of each edge.

Paths

• The Paths table contains lists of the nodes in a path queried by all users.

StartingPointsLog

• The StartingPointsLog table contains a list of starting positions that are 0.0001 degrees away from their closest edges/nodes.

InsertNode

• The file insertNode.php takes several inputs like Latitude, Longitude, Description, Building, RoomNumber and such attributes of a node and store them in the database as a new node.