Estimation of MineRo's Kinematic Parameters for Underwater Navigation Algorithm

Tae–Kyeong Yeu, Suk-Min Yoon, Soung-Jea Park, Sup Hong, Jong-Su Choi, Hyung-Woo Kim, Dae‐Won Kim, Chang-Ho LeeOcean 2011 · DOI

Abstract

A test miner named MineRo was constructed for the purpose of shallow water test of mining performance. In June of 2009, the performance test was conducted in depth of 100 m, 5 km away from Hupo-port (Korean East Sea), to assess if the developed system is able to collect and lift manganese nodules from seafloor. In August of 2010, in-situ test of automatic path tracking control of MineRo was performed in depth of 120 m at the same site. For path tracking control, a localization algorithm determining MineRo's position on seabed is prerequisite. This study proposes an improved underwater navigation algorithm through estimation of MineRo's kinematic parameters. In general, the kinematic parameters such as track slips and slip angle are indirectly calculated using the position data from USBL (Ultra-Short Base Line) system and heading data from gyro sensors. However, the obtained data values are likely to be different from the real values, primarily due to the random noise of position data. The aim of this study is to enhance the reliability of the algorithm by measuring kinematic parameters, track slips and slip angle.

Introduction: The Challenge of Underwater Navigation

Navigating a heavy tracked vehicle on the soft seafloor is a constant battle against "slip"—where tracks spin without moving the robot forward—and "drift." Underwater navigation is notoriously difficult because GPS signals cannot penetrate seawater. Researchers typically rely on Ultra-Short Base Line (USBL) systems, which use acoustic pings to triangulate a vehicle's position. However, USBL data is inherently "noisy"; factors like water temperature, salinity, and depth cause the reported position to jump around erratically, as shown in the raw data plots.

Raw position data from the USBL system showing significant acoustic noise.

For a tracked vehicle like MineRo (a 1/20 scale test miner), designed to collect manganese nodules from the seabed, precise positioning is critical for path tracking. Standard navigation filters, such as the Extended Kalman Filter (EKF), typically attempt to estimate "track slip" (sliding) and "slip angle" (the difference between heading and actual movement direction) indirectly from noisy USBL data. This often results in jagged, unreliable paths that hinder autonomous control.

Methods: Enhanced Kinematic Modeling

This paper introduces an enhanced navigation algorithm for the MineRo underwater miner that directly calculates kinematic parameters using onboard sensors, bypassing the unreliability of indirect estimation from noisy external sources.

Kinematic Modeling and Sensor Offset

The researchers utilized a mathematical model for tracked vehicles that accounts for the velocities of the inner and outer tracks, the heading angle, and the slip parameters. A key technical challenge is that the Doppler Velocity Logger (DVL) sensor is not located at the robot's center of mass (CM). To address this, the authors derived specific geometric transformations to translate the DVL's local velocity readings into the CM's velocity.

Kinematic model of MineRo showing the geometric relationship between the Center of Mass (CM) and the Doppler Velocity Logger (DVL).

Direct Parameter Estimation

Instead of treating slip ($i$) and slip angle ($\alpha$) as hidden states for the EKF to guess, the system uses a DVL and an Attitude and Heading Reference System (AHRS) to measure them in real-time. These directly calculated parameters are fed into the EKF's transition matrix. This allows the filter to rely on stable, high-frequency internal sensors for movement logic while using the noisy USBL data only as a global "sanity check."

Experimental Results and Key Insights

The study utilized real-world experimental data collected during sea trials in August 2010 near Hupo-port, East Sea of Korea, at a depth of 120 meters.

Path Smoothing and Reliability

The enhanced EKF approach significantly smoothed the navigation path. While raw USBL data and standard EKF outputs were jagged and discontinuous, the proposed method maintained a physically plausible, continuous trajectory. By decoupling slip estimation from noisy position data, the navigation system maintained stability even when the USBL signal was particularly noisy or momentarily lost.

Slip Quantification

The trials provided concrete measurements of the vehicle's interaction with the seabed:

  • Slip Angle: Measured up to 0.25 radians (~14 degrees).
  • Track Slip: Ranged from 0.1 during straight-line movement to 0.5 during aggressive turns.

The "aha moment" of this research is the realization that high-frequency internal sensors (DVL/AHRS) are far more trustworthy for local kinematics than external acoustic systems. By calculating the slip angle as the inverse tangent of lateral vs. longitudinal velocity, the robot gains a much better "feel" for its interaction with the soft sediment.

Conclusion and Future Work

While this kinematic model is highly effective for shallow water tests, several challenges remain for full-scale deployment. The ultimate goal for MineRo is operation at depths of 5,000 meters. At these extreme depths, acoustic noise and sensor latency will likely increase, requiring further tuning of the filter. Additionally, the current model assumes relatively uniform seabed consistency; future iterations may need to incorporate a full dynamic model accounting for variable soil mechanics to further improve accuracy.

Summary updated

Abstract

A test miner named MineRo was constructed for the purpose of shallow water test of mining performance. In June of 2009, the performance test was conducted in depth of 100 m, 5 km away from Hupo-port (Korean East Sea), to assess if the developed system is able to collect and lift manganese nodules from seafloor. In August of 2010, in-situ test of automatic path tracking control of MineRo was performed in depth of 120 m at the same site. For path tracking control, a localization algorithm determining MineRo's position on seabed is prerequisite. This study proposes an improved underwater navigation algorithm through estimation of MineRo's kinematic parameters. In general, the kinematic parameters such as track slips and slip angle are indirectly calculated using the position data from USBL (Ultra-Short Base Line) system and heading data from gyro sensors. However, the obtained data values are likely to be different from the real values, primarily due to the random noise of position data. The aim of this study is to enhance the reliability of the algorithm by measuring kinematic parameters, track slips and slip angle.
BibTeX
@inproceedings{yeu2011estimation,
  author = {Tae–Kyeong Yeu and Suk-Min Yoon and Soung-Jea Park and Sup Hong and Jong-Su Choi and Hyung-Woo Kim and Dae‐Won Kim and Chang-Ho Lee},
  title = {{Estimation of MineRo's Kinematic Parameters for Underwater Navigation Algorithm}},
  booktitle = {Ocean},
  year = {2011},
  doi = {10.4217/opr.2011.33.1.069},
}
Figure