Using a Distance Sensor to Detect Deviations in a Planar Surface

Carter Sifferman,William Sun,Mohit Gupta,Michael Gleicher

2024-08-07

Abstract:We investigate methods for determining if a planar surface contains geometric deviations (e.g., protrusions, objects, divots, or cliffs) using only an instantaneous measurement from a miniature optical time-of-flight sensor. The key to our method is to utilize the entirety of information encoded in raw time-of-flight data captured by off-the-shelf distance sensors. We provide an analysis of the problem in which we identify the key ambiguity between geometry and surface photometrics. To overcome this challenging ambiguity, we fit a Gaussian mixture model to a small dataset of planar surface measurements. This model implicitly captures the expected geometry and distribution of photometrics of the planar surface and is used to identify measurements that are likely to contain deviations. We characterize our method on a variety of surfaces and planar deviations across a range of scenarios. We find that our method utilizing raw time-of-flight data outperforms baselines which use only derived distance estimates. We build an example application in which our method enables mobile robot obstacle and cliff avoidance over a wide field-of-view.

Robotics,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper aims to address the problem of detecting geometric deviations (such as protrusions, objects, depressions, or cliffs) on planar surfaces using miniature optical Time-of-Flight (ToF) distance sensors. To achieve this goal, the paper proposes a method that leverages all the information contained in the raw time-of-flight data captured by commercial ToF sensors.

Specifically, the main contributions of the paper include:
1. **Problem Analysis**: The authors analyze the problem of detecting planar surface deviations using distance sensors and identify the main challenge—the ambiguity between geometry and surface photometry.
2. **Method Introduction**: A lightweight computational method is proposed to detect these deviations, utilizing the raw time-of-flight information from the sensor.
3. **Performance Evaluation**: The sensitivity and accuracy of the proposed method are characterized and compared with methods that rely solely on sensor distance estimates.
4. **Application Example**: The method is demonstrated in the context of obstacle avoidance for mobile robots.

The method in the paper assumes that the sensor's position relative to the planar surface remains constant (i.e., only translations parallel to the surface are allowed) and requires a small dataset to fit the surface model. The method is not intended to precisely identify the nature of the deviations but rather to classify the measurements as "planar" or "non-planar."

To overcome the ambiguity between geometry and photometry, the paper employs a Gaussian Mixture Model (GMM) to fit a small portion of the planar surface measurement data. This model implicitly captures the expected geometric shape and photometric distribution and is used to identify measurements that may contain deviations. Experimental results show that the method using raw time-of-flight data performs better than methods relying solely on distance estimates.

Additionally, the paper discusses related work, including obstacle detection methods based on machine motion and parallel ground, stereo vision methods, and learning-based methods, and highlights the differences between these methods and the proposed method. Finally, a series of experiments validate the effectiveness of the proposed method, particularly in obstacle detection performance at different distances.