Soldier Activity Recognition
- Resource Type
- Conference
- Authors
- Zhang, Andrew; Ebling, Maria R.
- Source
- 2023 IEEE MIT Undergraduate Research Technology Conference (URTC) Undergraduate Research Technology Conference (URTC), 2023 IEEE MIT. :1-5 Oct, 2023
- Subject
- Aerospace
Bioengineering
Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
General Topics for Engineers
Power, Energy and Industry Applications
Robotics and Control Systems
Training
Accelerometers
Lasers
Military vehicles
Activity recognition
Predictive models
Task analysis
activity recognition
long short-term memory
re-current neural networks
wearable
- Language
For years, the United States Army used the Multiple Integrated Laser Engagement System to allow units to emulate firefights during training. However, this system is outdated and difficult to use. A replacement system, under development, faces difficulty differentiating soldiers riding in vehicles from those marching behind. This paper assesses a system that utilizes accelerometers in Trigger Detection Modules to differentiate these activities. We collected data from soldiers while performing tasks such as “rucking” and riding in military and passenger vehicles. We added our data to the WISDM dataset and produced a predictive model for soldier activity recognition. Our model is over 97% accurate in differentiating between nine activities. It extends the existing ability to recognize everyday activities to include common soldier activities while requiring only an accelerometer. Our work presents the military with an innovative model that could be used to improve its training system.