Obstacle shape determination by mobile robot sensor system using GP2Y0A (Sharp) type IR distance sensors

Abstract

PURPOSE. When organizing a mobile robot (MR) movement in a non-deterministic environment, the SLAM problem arises, which includes the detection of an obstacle presence by the MR sensor system, the distance to the obstacle and its shape. To solve this problem, an infrared (IR) analog distance sensor is often used, the information flow from which is relatively small and can be processed in real time using low-performance microcontrollers. However, such a sensor can only detect an obstacle and determine а distance to a certain point on its surface. The goal is to develop a method for determining both а distance to an obstacle and its shape. When setting up experiments on the use of an analog sensor of the GP2Y0A (SHARP) type, a problem was revealed associated with the occurrence of not only fluctuation noise in а data communication channel, but also artifacts – anomalous signal values appear with random periodicity. It is necessary to determine the source of such interference, propose a method for estimating its parameters and a way to minimize its influence. METHODS. To determine the shape of an obstacle, a differential method is proposed based on the use of several readings of a scanning IR distance sensor. As an indicator of the “noisiness” of the channel, it is proposed to use the number of sensor signal values that exceed the average signal value by 1σ, 2σ, 3σ, 4σ and 5σ. The use of standard methods for filtering abnormal values of a sensor signal leads to significant delays in a response of the MR control system. This is unacceptable, because at the executive level of a control system it is required to provide a "hard" real-time mode. RESULTS. The article presents the results of experiments showing the conditions for applying the differential method for determining a shape of an obstacle, the source of anomalous signal values is identified and a method for minimizing them is proposed. A method for increasing the practical use range of a nonlinear IR sensor conversion function is also proposed. CONCLUSION. The number and magnitude of abnormal values depend on a communication channel length. When using analog sensors, it is necessary to convert an output signal into digital form using analog-to-digital converters (ADCs) in an integrated design, structurally bringing the ADC as close as possible to the signal source.

Keywords

• mobile robotics
• control system
• analog sensor
• communication channel
• fluctuation noise
• noise filtering