Bio-Inspired Motion Sensor Simulation for Advanced Robotic Navigation

Introduction

Animals are led to develop many faculties to evolve in their environments. That is why we are interested in their capabilities and why man is seeking to reproduce them in technology. The calculation of optical flow is one of the main projects. This calculation would help robots avoid obstacles. Indeed, photodiode sensors would obtain the light contrast information that will be processed to obtain the optical flux.

Theoretical Framework

The purpose of this project is to analyze the data collected by two photodiode sensors offset by an angle ∆φ =3.

8 °. We will therefore process the perceived signal in order to calculate the time offset of the two sensors between two contrast changes. Once the time offset has been calculated, the optical flux is calculated using the equation :

𝜔 = ∆𝜑 ∆𝑡

Data Acquisition and Processing

The data is captured by two photodiode sensors called LSC. They are offset by an angle ∆𝜑 = 3,8 ° and have a sampling frequency of 𝑓𝑒 = 2𝑘𝐻𝑧. We can see on the picture below different boxes, these boxes represent the sensors, we will only use two of them.

Get quality help now

KarrieWrites

Verified writer

Proficient in: Technology

5 (339)

“ KarrieWrites did such a phenomenal job on this assignment! He completed it prior to its deadline and was thorough and informative. ”

+84 relevant experts are online

Hire writer

Filtrage

The signal at the output of the sensors is noisy, so it must pass through a low-pass filter with a cut-off frequency 𝑓𝑐 = 30 𝐻𝑧 (to keep only the information we are interested in). In addition, we must subtract the DC component from the signal that is not useful to us in our analysis.

Absolute alue

The absolute value of the signals at the filter output is realized in order to facilitate the thresholding calculation step (calculation of contrast differences).

Because what interests us is the difference in contrast, no matter if we go from zero to negative or zero to positive, it comes back to the same thing.

Threshold setting

The purpose of this step is to analyze when (through the data) there is a change in contrast. This step is a key step in the project. Because threshold values if they are too large, Photodiode LSC

Simulation of a bio-inspired motion sensor destroy the information while threshold values too low add false information. We have two curves with binary values at the end of this step.

Calculation of the time lag

The objective of this step is to analyze the information coming out of the thresholding. The time difference between a contrast difference from the first sensor and the same contrast difference from the second sensor must be calculated. We therefore have at the end of this step a single curve grouping all the time lags over time.

Calculation of the optical flow

Using the formula quoted at the beginning of the section, we associate a time shift with an optical flow and we therefore have at the end of this section a curve with the optical flow, information we were looking for.

From Theory to Practice Using Matlab and Simulink

Data acquisition using Simulink

To analyze the information using Simulink you need to import the data (which was in Matlab's Workspace). For this reason we used the Fromworspace block followed by a multiplexer to associate the data on the same curve and facilitate the continuation of the different operations.

Filtrage

For the filtering we first used a loop in which there was a Mean block to remove the DC component of the signal (Fig. 2). Then we used Matlab's Butter function, which allows us to choose the order of the filter (here 4) and to enter our cutoff frequency associated with our sampling frequency (2𝑓𝑐𝑓𝑒÷2). At the output of this function we have the numerator and denominator to enter into the Simulink Discrete Filter block (Fig. 2) after the multiplexer to perform our filtering.

Absolute value and threshold setting

After having realized an absolute value using the Abs block we obtained this curve: Now comes the thresholding operation, for this we use the hysteresis comparator modeled with the Relay block in Simulink. We have chosen to use switch on 75 and switch off 30 as the values (these values allow us to have better results).

Calculation of the time lag

To achieve the time shift we associated these block diagrams (Fig. 3) with a Counter Free-Running block, which allowed us to calculate the time difference between two rising edges of the two photos.

Calculation of the optical flow and result

A curve was previously made to associate a time shift with an optical flow. Using a Direct Lookup Table block (n-D) we resulted in the following: In orange is the gyrometer data and in blue is our analyzed data.

Results and Conclusion

The simulation results, when compared with gyrometer data, demonstrated the potential accuracy of this bio-inspired approach to calculating optical flow. While there were discrepancies likely due to sensor data acquisition and the thresholding process, the project highlights the viability of bio-inspired motion sensors in robotics. The work of Lacluque and Levain not only contributes to the field of bio-robotics but also underscores the importance of interdisciplinary approaches in advancing technology.

In conclusion, the simulation of a bio-inspired motion sensor represents a promising step toward the development of more sophisticated and efficient robotic systems capable of navigating complex environments. This project not only showcases the potential of bio-inspired technologies but also paves the way for further exploration in the convergence of biology and robotics.