Performance Evaluation and Error Mitigation of Ultrasonic Indoor Positioning: An ESP32-Based IMU-ESKF Architecture

Area of Science:

• Robotics and Automation
• Indoor Localization Systems
• Sensor Fusion

Background:

• Reliable indoor localization is critical for automated guided vehicles (AGVs), robot validation, and industrial digital twins.
• Ultrasonic positioning systems suffer performance degradation due to changing acoustic visibility, multipath effects, and NLoS (non-line-of-sight) conditions.

Purpose of the Study:

• To evaluate the robustness of Marvelmind Super-Beacon localization under varying conditions.
• To develop and assess an embedded ultrasonic-inertial pipeline for mitigating acoustic positioning failures.
• To validate the proposed mitigation framework across different Marvelmind deployment architectures (NIA and IA).

Main Methods:

• Implemented an embedded ultrasonic-inertial pipeline on an ESP32-S3-WROOM-1 module.
• Combined UART packet validation, high-frequency inertial acquisition (ICM-20948 at 500 Hz), outlier rejection, and a 15-state error-state Kalman filter (ESKF).
• Quantitatively assessed mitigation performance using firmware-consistent replay of recorded AGV and UR10 datasets.

Main Results:

• Replay-based trial-mean RMSE for 2D AGV localization decreased from 101.2-104.1 mm (raw ultrasonic) to 47.2-48.7 mm (fused data).
• Peak failure-interval errors in AGV scenarios were reduced by 64.2-65.7%.
• For 3D UR10 robot-arm positioning, replay-based trial-mean RMSE decreased from 157.6-158.4 mm to 80.2-80.5 mm, with peak 3D errors reduced by 58.8-60.0%.

Conclusions:

• Demonstrated the feasibility of embedded ultrasonic-inertial fusion for enhancing localization robustness in controlled laboratory AGV and robot-arm scenarios.
• The proposed mitigation framework effectively reduces positioning errors and improves continuity during acoustic signal degradation.
• Further validation is required for large-scale, dynamic industrial environments; closed-loop control based on fused data is a future research direction.