Ruggedized Ultrasound Sensing in Harsh Conditions: eRTIS in the wild

TL;DR

This paper addresses the challenge of robust sensing in harsh industrial environments where optical systems falter by introducing eRTIS, a fully embedded, GPU-accelerated broadband ultrasound sensor. It combines a 32-element MEMS microphone array with a broadband capacitive transducer and log-FM chirps, enabling 2D/3D beamforming through CUDA-accelerated processing on NVIDIA Jetson hardware. The design emphasizes modular hardware, a rugged IP-rated enclosure, external and in-band synchronization, and a flexible software stack for single- and multi-sensor deployments, achieving real-time acoustic imaging at low latency. Real-world evaluations in harbor, off-road, and indoor cluttered settings demonstrate that eRTIS provides robust sensing where optical systems degrade, and can integrate with ROS-driven pipelines for robotic perception and navigation.

Abstract

We present eRTIS, a rugged, embedded ultrasound sensing system for use in harsh industrial environments. The system features a broadband capacitive transducer and a 32-element MEMS microphone array capable of 2D and 3D beamforming. A modular hardware architecture separates sensing and processing tasks: a high-performance microcontroller handles excitation signal generation and data acquisition, while an NVIDIA Jetson module performs GPU-accelerated signal processing. eRTIS supports external synchronization via a custom controller that powers and coordinates up to six devices, either simultaneously or in a defined sequence. Additional synchronization options include bidirectional triggering and in-band signal injection. A sealed, anodized aluminum enclosure with passive cooling and IP-rated connectors ensures reliability in challenging conditions. Performance is demonstrated in three field scenarios: harbor mooring, off-road robotics, and autonomous navigation in cluttered environments, demonstrates that eRTIS provides robust sensing in situations where optical systems degrade.

Figures (11)

• Figure 1: a) Shows the schematic representation of the hardware architecture showing the front and back-end with their respective subsystems and interconnections. b) The eRTIS sensor printed circuit boards with the randomized 32-element microphone array on the left and the 6×5 regular grid front-end in the middle. The back-end PCB is shown on the right with an NVIDIA Xavier NX system on module inserted in the DDR4 socket. In this sub-figure the following components are highlighted 1. the 32-element MEMS microphone array 2. the 180 V and 200 V step-up regulation circuits 3. the Senscomp 7000 ultrasound transducer 4. the two-stage gain amplifiers that amplify the output signal 5. the connectors for providing power and interfacing with the sensor 6. the NVIDIA System on Module 7. USB connectors for providing extra data storage capabilities or directly interfacing with the microcontroller 8. one of two CSI connectors that provide a means for connecting a camera directly with the NVIDIA SoM.
• Figure 2: The eRTIS anodized aluminum case is shown displaying it's ingress proof connectors on the left side that provide power and interfaces for data transfers and triggering the sensor. In the middle the front of the case is showing the black mesh membrane through the cutouts in the enclosure and on the right is the back of the case showing the cooling slots and two status light indicators.
• Figure 3: An exploded view of the CAD model of the eRTIS sensor and its anodized aluminum enclosure together with its seals and mesh membrane for dust and water ingress protection.
• Figure 4: In order to verify the heat dissipating properties of the designed enclosure, a stress test was devised where the device was placed in an ambient office environment and both the CPU and GPU of the NVIDIA Jetson Nano were put under a constant synthetic load for over 2 hours of constant operation. The left plot shows the temperature of the CPU, GPU and PLL of the NVIDIA Jetson Nano whereas the two plots on the right show the respective load of the GPU and CPU cores.
• Figure 5: Comparison of the received ultrasound frequency spectrum with and without the protective mesh membrane.