Applied Modeling Techniques and Data Analysis 1

6 Error Detection in Sequential Laser Sensor Input

In this chapter, we build two low complexity and robust error-detection algorithms that adjust for sensor faults and inaccuracies in real time. The algorithms can detect consecutive and abrupt sensor slips and provide a paradigm for real-time corrections. In addition to reliability, the algorithms do not require prior data knowledge, nor do they rely on any assumptions about the distribution of the data. The run time is independent of the input size, which is ideal for large volumes of data, and allows for implementation at the sensor level.

6.1. Introduction

A major obstacle to progress in robotics is a machine’s limited ability to interact with the environment. Fortunately, the cost of laser sensors has decreased, making them an attractive choice as detection tools. Laser sensors are low-cost devices that are readily available and simple to operate. However, laser sensors can suffer from jitter and are not very reliable, especially when they encounter reflective surfaces. The trade-off for their convenience is the rudimentary quality of the data received, which explains the motivation for research on the topic of increasing sensor accuracy.

Prior research on sensor accuracy has been in either engineering or data processing. Papers in the field of engineering focus on the physical aspects of sensor construction and data acquisition, with a goal of obtaining better data through hardware improvements (see, for example, Shirafuji ...