Prediction of Human Walking Trajectory Using Stochastic Path Planning Model Through UAV
Our proposal is to design and build a model used for detecting and accurately predicting pedestrian activity using scalable stochastic algorithms and convert the approximate real-time coordinates of the pedestrians into localized coordinates (in pixels) for effectively calculating free-flow velocity
2025-06-28 16:28:51 - Adil Khan
Prediction of Human Walking Trajectory Using Stochastic Path Planning Model Through UAV
Project Area of Specialization Artificial IntelligenceProject SummaryOur proposal is to design and build a model used for detecting and accurately predicting pedestrian activity using scalable stochastic algorithms and convert the approximate real-time coordinates of the pedestrians into localized coordinates (in pixels) for effectively calculating free-flow velocity of pedestrians using UAVs. The primary objective for this model is to detect and predict human walking trajectory is to ensure safety and comfort for people, to mitigate risk in dense crowds. Buildings and events can be planned with a focus on safety, making use of virtual experience gained from pedestrian streams simulation and short-term prediction of real-life scenarios. Short term simulation predictions could be used to warn of critical situations such as danger of high densities.
Project Objectives- To gather information for the purpose of selecting the appropriate equipment (camera) that can capture aerial images from a certain height.
- To gather and manage datasets for different groups of pedestrians at different angles and positions.
- To use a machine learning algorithm for training our model so that it can enable it to detect humans from the scenery which may contain a vast number of objects.
- To identify and choose which detection technique will work best with our model which is computationally efficient.
- To obtain parameters for the stochastic path planning model such as distance between pedestrians, free-flow velocities etc. from the dataset
- To implement the suitable stochastic path planning algorithm and predict the future trajectory of the walking human as well as his final position by determining their final coordinates.
- A handheld device will display their predicted trajectory and final position.
- Aerial videos captured via UAV having default coordinates.
- Human detection in image using machine learning algorithm
- Calculating real-world parameters such as free-flow velocities and crowd density.
- Calibrating the human motion model using real-world parameters.
- Human walking trajectory prediction using the obtained coordinates.
- Obtaining and displaying the proposed trajectory on an HMI.
- Spying on potential suspects during Military operations.
- Public safety and traffic control.
- Can be used by municipalities to mitigate risk in dense crowds such as crowded railway stations
- Buildings and events can be planned with a focus on safety, making use of virtual experience gained from pedestrian motion simulations.
- Can also be used by autonomous car manufacturers to predict the potential trajectory of pedestrians hence preventing road accidents.
The UAV is controlled both manually and autonomously to cope with worst case scenarios. Our project can be divided into following major parts:
1. Human Detection:
In the first step, aerial images and videos along with the GPS Coordinates from the image processing system installed in the UAV are captured.
2. Human Walking Trajectory Prediction:
Flight motion and image variables are provided to the visual tracking and geolocation module so that the module can use these parameters to predict the potential trajectory and future position of the target. UAV real time location is obtained using flight motion.
3.Obtaining GPS Coordinates from Target Coordinates:
Future localized coordinates of the target are predicted using the stochastic path planning algorithm and these are converted into GPS coordinates using a transform that converts these localized coordinates into GPS coordinates.
Final Deliverable of the Project HW/SW integrated systemCore Industry SecurityOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Peace and Justice Strong InstitutionsRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 69500 | |||
| Nvidia Jetson Nano (Processor) | Equipment | 1 | 19500 | 19500 |
| GoPro Hero 6 (Camera) | Equipment | 1 | 50000 | 50000 |