https://github.com/sid220/shot-prediction

A neural-network based approach to predicting the binary outcome (hit, miss) of shooting an object into a hole or at a target.
https://github.com/sid220/shot-prediction

Last synced: 4 days ago
JSON representation

A neural-network based approach to predicting the binary outcome (hit, miss) of shooting an object into a hole or at a target.

• Host: GitHub
• URL: https://github.com/sid220/shot-prediction
• Owner: Sid220
• License: mit
• Created: 2024-03-06T21:24:50.000Z (9 months ago)
• Default Branch: master
• Last Pushed: 2024-03-06T21:31:16.000Z (9 months ago)
• Last Synced: 2024-10-11T16:50:59.753Z (about 1 month ago)
• Language: Jupyter Notebook
• Homepage:
• Size: 551 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Shot Prediction

[![dataset](https://huggingface.co/datasets/huggingface/badges/resolve/main/dataset-on-hf-sm.svg)](https://huggingface.co/datasets/sid220/2713-2024-shot-prediction/blob/main/raycasting_normalised_angles.csv)

A neural-network based approach to predicting the binary outcome (hit, miss) of shooting an object into a hole or at a

target.

## Introduction

This project provides a neural-network based approach to predicting the binary outcome (in, miss) of shooting an object

into a hole or at a target. The model is trained on a dataset of shots and their outcomes.

## Simulation

The notebook `data_gen.ipynb` contains a simulation of a shooting process. The simulation is based on a linear shot, and

uses raycasting.

Here's an example of the simulation (red = made, blue points are corners of a rectangle representing the goal):

![Simulated shots](other/sim.png)

You can grab the data

from [https://huggingface.co/datasets/sid220/2713-2024-shot-prediction/blob/main/raycasting_normalised_angles.csv](https://huggingface.co/datasets/sid220/2713-2024-shot-prediction/blob/main/raycasting_normalised_angles.csv).

## Usage: FRC team 2713's 2024 robot



FRC Team 2713, [Red Hawk Robotics](https://www.thebluealliance.com/team/2713), uses this project to predict the outcomes

of shooting orange rings, or "notes", on their 2024 robot. Their robot includes a pivot shooter and chassis which

functions as a turret.



Data from the robot's sensors will be used as input to the model (later annotated as to whether the shot went in), and

the

model's output will be used to control the robot's position and velocity to make the "optimal shot". For now, data

is [simulated](#Simulation). You can grab the data

from [https://huggingface.co/datasets/sid220/2713-2024-shot-prediction/blob/main/raycasting_normalised_angles.csv](https://huggingface.co/datasets/sid220/2713-2024-shot-prediction/blob/main/raycasting_normalised_angles.csv).

Here's the output of the model when the x and y are the x and y of the point on the graph, the shot is off the horizon 45 degrees, the chassis is slightly pointing right, and the velocity is 0

with the red representing the rough location of the goal:

![Model output](other/predictions.png)

Training is fairly simple, and can be found in `train.ipynb`.