The work in this paper describes implementation of Artificial Neural Network (ANN) on space processor LEON3. The ANN has been tested for training voice signal and for detecting anomaly signal on multiple analog sensors. The build-in radiation hardened UART 115200 interface of standard Space Hardware was utilized to receive compressed data from Artificial Intelligence (AI) Kit. The AI Kit was built to acquire human voice (as voice command) or to process input signals from multiple sensors concurrently. The Kit enables voice intuitively by pressing a training button and selecting proper command type from the keypad. The experiment results show that voice commands were detected successfully with accuracy of more than 95%. The second experiment was carried out by using analog sensor signals mixer to allow AI to learn and determine type of sensor data anomalies when some failures occur. The anomalies types were generated by adding unexpected stimulus signals to AI Kit analog input terminals. The result shows that the anomalies can be detected with accuracy of 80%. The size of AI Kit is relatively small and it was built with commercial components that enable replacement with space radiation hardened components. The space computer platform is based on LEON3 processor core and synthesized on Xilinx Virtex-5QV Field Programmable Logic Arrays (FPGA). The core runs at 100 MHz. The feed forward artificial neural networks Algorithm was implemented on Real-Time Executive for Multiprocessor Systems (RTEMS) operating system. The AI Kit consists of audio signal pre-amplifier, automatic gain control circuit, analog signal buffer circuit, high pass filter circuit (HPF), analog to digital converter ADC which is integrated in the 8 bit microcontroller. The modified FFT algorithm that runs on microcontroller is used for data compression and for increasing uniqueness of the data acquired. A DC/DC converter for battery usage is included, when 5V voltage supply is not available.
Artificial Intelligence Implementation on Voice Command and Sensor Anomaly Detection for Enhancing Human Habitation in Space Mission
2019-06-01
899070 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
British Library Conference Proceedings | 2000
|SAE Technical Papers | 2001
|Conceptual design of a habitation module for a deep space exploration mission
Tema Archiv | 2013
|Conceptual design of a habitation module for a deep space exploration mission
Online Contents | 2013
|