Simulink Student Challenge Winners
MathWorks announces the winners of the 2023 Simulink Student Challenge. Congratulations and thanks to all the students who entered.
Dynamic simulation of a pendulum-driven spherical rover
Politecnico di Torino, Italy – Vincenzo Cicero
This project prototypes a pendulum-driven spherical rover designed in Simulink. The model consists of two opposing motors that provide propulsion for forward and backward motion. The pendulum, present within the motor shaft, counteracts the torque applied by the motors and provides steering ability by controlling the inclination of the pendulum.
The ASTRA team models the physical system for the rover using Simscape. They design and implement a sliding mode controller and a linear quadratic regulator to effectively control the two main motion directions. The team does an excellent job of presenting and explaining the design, analysis, and results of their model. They achieve their overall objective of enabling the rover to adapt to different conditions and to reach a desired position in space within a certain time frame.
This innovative project is a great example of how Simulink can be used for rapidly prototyping systems and validating their performance. The overall presentation is engaging, clear, and informative. We wish the ASTRA team good luck on their incredible mission of space exploration!
Real-time prediction and feedback of grasp-types of a prosthetic using LSTM
Egypt-Japan University of Science and Technology, Egypt – Sudhir Solomon Zhuwawu
This project uses MATLAB and Simulink to enable prosthetic hand users to know how they are grasping an object without needing to look at the hand using a multi-channel haptic feedback device.
Sudhir used the Simulink Support Package for Arduino Hardware to collect sensor data via Simulink from a prosthetic hand and used the data to train and test a LSTM network that he created using the Deep Learning Toolbox. He then used the trained network in his Simulink model to make continuous real-time grasp type predictions of the prosthetic hand that then outputs commands to a third-party 3-D mechanical system simulator. These commands update the 3-D simulation with haptic-feedback stimulation patterns that correspond to different grasp types in real-time.
Overall, this project is an exemplary example of how Simulink can be used to take a novel concept that started off as just an idea, to a prototype system that can be used to control 3D simulations in real-time with minimal time and resources. We wish Sudhir the best of luck on his journey to improve the accessibility of prosthetic hands and we are excited to see what is yet to come out of this project!
King’s College London, England – Matteo Liguori
This project shows the simulation of a self-balancing bike, employing a controller to enable active balancing and the ability to complete waypoint missions, along with the ability to employ a variety of different sensor fusion algorithms including the Kalman filter and modelled sensor error.
Matteo developed a closed-loop controller for a self-balancing bicycle that can autonomously navigate towards waypoints. The system was designed to operate using noisy wheel encoders and IMU sensors, which were used to estimate the bicycle's state in real-time. The experimental results showed that the sensor fusion approach was effective in improving the accuracy and reliability of the state estimation process.
Overall, the project demonstrates how MATLAB and Simulink can be used to build comparative analysis of sensor fusion techniques for state estimation.