This week I’ve mainly been working on figuring out the most efficient and effective way to present our pathfinding algorithm. I have experimenting with the best method to record my portion of the final presentation video. Ideally, slides would help as a visual guide and I would be voicing over to help explain the diagrams. Finally, I have also been working on the final report. This will also be the majority of work left for the following week.
This week I majorly worked in integrating my path planning algorithm with Zeyi’s user interface. In the process, we identified some bugs with the algorithm and I fixed them. First bug was with the replan() function. I ended up referencing some local variable that only existed in my version of the path planning algorithm (duplicate variable names in testing code). This created a problem on Zeyi’s side where replan() only works for certain pairs of nodes. It was a rather…
Changed the map of the shopping to the form of “multiple items on a single node”. Formatted the interface’s cart-guiding page by adding a picture of the map to it. Changed the Node List and Item List to display the remaining nodes and items on the planned route. Implemented all functions of the simulation buttons. To-do: further revise the look of the pages testing the interface for bug
Finished Integrating the user interface with the algorithm for the final demo. Added buttons for simulations of finding obstacles, and implemented functions of skipping and placing items. Next step: Furnishing the cart tracking page for the final presentation, seperate the simulation and tracking components.
This week, I first addressed the problem with local re-planning. I was thinking if an obstacle were to appear in the optimal route between Node1 and Node2, I can simple recall the saved shortest path between Node1 and Node2 that has been pre-planned to “go around” the obstacle. But I soon realized that due to the way my algorithm is set up, every pair of adjacent nodes in the optimal route are neighbors. This leads to a problem that if…
This week I worked on setting up the demo and integrating the range sensor code with Zeyi and Zehong’s pipelines. I also worked on the viability of altering our design to include two forward facing range sensors. To measure the viability of altering our design to include 2 forward facing sensors, I decided to create a model of what a single range sensor is capable of doing. This involved doing the trigonometry to solve for the distance of the field…
Updated some features of the user interface’s second page. Added a button for the user to skip the current item. Implemented the feature that the user would be notified when the cart arrives at the current item, and an event loop that simulates the process that the cart travels on the map. The next step of the user interface part will be integrating with the path planning algorithm and refining the look of the interface.
After talking with professor during the last meeting, I first updated the algorithm so that it can start and end of different nodes. I then embarked on testing the limits of my algorithm. To do that, I first need to generate maps of my own. I first randomly generate the x and y coordinates of n nodes. The weight(distance) of edge between each pair of nodes is always the euclidean distance. I chose this to be the distance to avoid problems…
In the past week I have been testing the detection ranges of the range sensor. I ran multiple instances of 2 experiments. The first experiment measures the accuracy of range sensor outputs from a known distance from the wall. This experiment was run multiple times by varying the distance from the wall and logging the sensor outputs to a spreadsheet. The 2nd experiment varies the angle that the sensor points at an object. I varied the angle from 3 degrees…
This week, I finished implementing the entire path planning algorithm. It takes in two inputs, an origin and a list of nodes the cart needs to visit; and it returns the optimal route. As promised, the algorithm pre-computes the shortest distance between each path of nodes in the map using A* with the heuristic function being the euclidean distance between the two nodes. The shortest paths between each path of nodes are stored in data structure to be used in…