Deep Learning Based Fall Recognition and Forecasting for Reconfigurable Stair-Accessing Service Robots

Jun Hua Ong,Abdullah Aamir Hayat,Braulio Felix Gomez,Mohan Rajesh Elara,Kristin Lee Wood

DOI: https://doi.org/10.3390/math12091312

IF: 2.4

2024-04-26

Mathematics

Abstract:This paper presents a comprehensive study on fall recognition and forecasting for reconfigurable stair-accessing robots by leveraging deep learning techniques. The proposed framework integrates machine learning algorithms and recurrent neural networks (RNNs), specifically Long Short-Term Memory (LSTM) and Bidirectional LSTM (BiLSTM), for fall detection of service robots on staircases. The reconfigurable stair-accessing robot sTetro serves as the platform, and the fall data required for training models are generated in a simulation environment. The two machine learning algorithms are compared and their effectiveness on the fall recognition task is reported. The results indicate that the BiLSTM model effectively classifies falls with a median categorical accuracy of 94.10% in simulation and 90.02% with limited experiments. Additionally, the BiLSTM model can be used for forecasting, which is practically valuable for making decisions well before the onset of a free fall. This study contributes insights into the design and implementation of fall detection systems for service robots used to navigate staircases through deep learning approaches. Our experimental and simulation data, along with the simulation steps, are available for reference and analysis via the shared link.

mathematics

What problem does this paper attempt to address?

The problem this paper attempts to address is the potential falling issue of stair cleaning service robots during task execution. Specifically, the paper focuses on recognizing and predicting the robot's falling behavior through deep learning techniques to ensure the robot's safety and prevent damage to the surrounding environment and personnel. The research focuses include:

1. **Fall Recognition**: Real-time detection of whether the robot has fallen through sensor data (such as IMU data).
2. **Fall Classification**: Classifying fall events into different types, such as falling backward, falling to the left, falling to the right, etc.
3. **Fall Prediction**: Predicting future fall events based on historical data and patterns to take preventive measures in advance.

Solving these problems is of great significance for improving the reliability and safety of stair cleaning robots, especially in complex and variable environments. Through this research, more intelligent and safer service robots can be developed, reducing the occurrence of accidents.