Electronic Systems Laboratory

Third Robotics Research Group Meeting with our speakers:
Andrew Murdoch – “A partial end-to-end reinforcement learning framework for autonomous racing.”
Autonomous racing is a rapidly growing research field linked to the increasing popularity of self-driving cars. To race competitively, vehicles must be controlled at their handling limits, spurring innovation in the field of autonomous driving. In this talk, we will see how current system architectures that leverage machine learning for vehicle control can be improved upon by incorporating ideas from the 'classic' autonomous driving pipeline.

Mohamed Zahier – “Automated Landing of a Fixed Wing Aircraft onto a Moving Platform”
The increased use cases of fixed-wing UAVs have created a demand for a system to automatically land the UAV in a safe and reliable manner. This project aims to meet this demand by designing, simulating, implementing, and practically testing flight and guidance control systems that can navigate the aircraft on a trajectory and land it on a moving platform. The flight control system consists of a hybrid architecture combining classical control with model predictive control.
https://youtu.be/pqp1b0lQDLE

Work on DockSat is still continuing. Looking forward to have some new progress and some demonstrations in the near future.

Pose estimation for CubeSat docking – Electrical and Electronic Engineering CubeSats are nanosatellites of standardised dimensions based on multiple 10-cm-sided cubes. These satellites play various roles in space including being used for both educational and technology-demonstration purposes. In future, they could also serve as the building blocks of other, larger missions....

We recently redid our website. Reducing overhead and simplifying the interface. Have a look.

Electronic Systems Laboratory The Electronic Systems Laboratory is a postgraduate research laboratory focusing on autonomous vehicles, housed within the Electrical and Electronic Engineering Department of Stellenbosch University, South-Africa. A Research Lab The core of the ESL is graduate students, primarily conducting research...

Sixteenth Robotics Research Group Meeting with our speaker:
Angel Gonzalez Llacer (MathWorks) – “Developing a Snake robot using an MBD approach”
Angel is Software Engineer at MathWorks who will be taking us through the full design process that he used to take his snake robot from an idea to development using MathWorks tools such as Simulink and Simscape. Angel will focus on the model-based design (MBD) workflow that he used to develop his robot through simulation. Angel will demonstrate workflows for sizing out hardware equipment such as motors and assessing the performance of the robot in different environments prior to purchasing any equipment. Angel will demonstrate the deployment of his algorithms and models onto a Arduino/ Raspberry Pi.

Robotics Research Group Meeting 16 Our speaker is:Angel Gonzalez Llacer (MathWorks) – “Developing a Snake robot using an MBD approach”Angel is Software Engineer at MathWorks who will be taking...

Fifteenth Robotics Research Group Meeting with our two speakers:
Darshan Makan and Leo Rocco Breedt (Opti-Num Solutions) – “Developing robotic applications with ROS and robotic system toolbox"
Robotic design can be a complex and expensive process that can be simplified and made cost effective through simulation and Model Based Design approaches. In this presentation we will take a look at the capabilities of the ROS and robotic system toolbox and some of the functions that can aid in path planning, collision checking and more. We will practically demonstrate examples by executing a pick and place workflow using a model of an ABB YuMi robot and simulate it using Simscape.

Robotics Research Group Meeting 15 Our speakers are:Darshan Makan and Leo Breedt (Opti-Num Solutions) – “Developing robotic applications with ROS and robotic system toolbox"Robotic design can ...

We had another successful balloon launch this past weekend from Hopefield in the Western Province. In the container were three final year projects and some electronics from a Masters project. The payloads were collected just outside of Porterville.

Fourteenth Robotics Research Group Meeting with our two speakers:
Darshan Makan and Kelishia Radhalal (Opti-Num Solutions) – “AI in Engineering"
AI is contributing to a significant transformation in the role machines play in our lives with techniques, like Machine Learning and Reinforcement Learning, rapidly being adopted to assess big data and unlock insights. Integrating AI into an engineering workflow can seem daunting and complex at first but we will explore how these capabilities are now accessible to engineers and scientists without requiring highly specialised expertise. During the seminar we will demonstrate the use of the MathWorks Reinforcement Learning Toolbox to develop and train an RL agent to optimize the control of a water distribution system.

Robotics Research Group Meeting 14 Our speakers are:Darshan Makan and Kelishia Radhalal (Opti-Num Solutions) – “AI in Engineering"AI is contributing to a significant transformation in the role...

Thirteenth Robotics Research Group Meeting with our two speakers:
Andrew Murdoch – “Developing a reinforcement learning agent to play Tetris”
Tetris is a popular testbed for algorithm development in the reinforcement learning community. Although there have been many successful attempts at creating agents to play the game with super-human performance, most of them have suffered from excessively long training times. This project investigates the use of a simple technique (model-free learning with linear function approximation) that results in faster training times.

Mr JC Schoeman – “Degenerate Gaussian factors for probabilistic inference”
In this paper, we propose a parametrised factor that enables inference on Gaussian networks where linear dependencies exist among the random variables. Our factor representation is a generalisation of traditional Gaussian parametrisations where the positive-definite constraint (of covariance and precision matrices) has been relaxed. For this purpose, we derive various statistical operations and results (such as marginalisation, multiplication and affine transformations of random variables) which extend the capabilities of Gaussian factors to these degenerate settings. By using this principled factor definition, degeneracies can be accommodated accurately and automatically at little additional computational cost. As illustration, we apply our methodology to a representative example involving recursive state estimation of cooperative mobile robots.

Robotics Research Group Meeting 13 Our two speakers are:Andrew Murdoch – “Developing a reinforcement learning agent to play Tetris”Tetris is a popular testbed for algorithm development in the ...

Twelfth Robotics Research Group Meeting with our two speakers:
Raynhardt Van Zyl – “Multi-Tier hardware for SLAM processing”
SLAM is computationally expensive, creating an issue for embedded devices on most terrestrial drones. The talk will be discussing a work around to break up processing into two major parts. Firstly, the computer vision, which is processed on the embedded device and secondly, the feature matching and batch processing of the feature map updates done on a remote PC.

Dr Willem Jordaan – “Controlling a Multirotor with a Suspended Payload”
In this talk we discuss a number of previous methods in which the problem of a multirotor transporting a suspended payload. Suspended payloads dramatically change the flight characteristics of the vehicle and needs to be compensated for. Techniques and results stretching over a number of previous and current research projects will be presented.

Robotics Research Group Meeting 12 Our two speakers are:Raynhardt Van Zyl – “Multi-Tier hardware for SLAM processing”SLAM is computationally expensive, creating an issue for embedded devices o...

Eleventh Robotics Research Group Meeting with our two speakers:
Annika Hohbach – “Trajectory Planning for a Rotary-wing UAV with an Unknown Payload”
Rotary-wing unmanned aerial vehicles are used in deliveries that might require complex manoeuvres, where a suspended payload induces unwanted oscillations that could cause the system to become unstable. Trajectory planning techniques depend on an accurate model of the system to ensure sufficient tracking to complete the desired task. A method to solve for minimum-swing trajectory needs to be developed that incorporates the unknown dynamics of the suspended payload.

Andries Burger – “Robust star detection and identification using a neural network”
Star trackers are used to determine the attitude of a satellite with high precision based on the stars detected and identified in the image. Star trackers use a full-sky star identification algorithm as they have no initial altitude, requiring the identification algorithm to be reliable for the star tracker to be able to enter its tracking mode. The aim of implementing a neural network is to improve on existing matching algorithms.

Robotics Research Group Meeting 11 Our two speakers are:Annika Hohbach – “Trajectory Planning for a Rotary-wing UAV with an Unknown Payload”Rotary-wing unmanned aerial vehicles are used in de...

Tenth Robotics Research Group Meeting with our two speakers:
Devlon Erasmus – “Trajectory Optimisation on a Bipedal Robot”
Legged robotics has become more prominent in recent years, led by the developments in trajectory optimisation methods. This work investigates trajectory optimisation methods for a pneumatically actuated bipedal robot. The aim will be building a bipedal robot and implementing different trajectories on it.

Nathan Weiss – “Design of a Bipedal Robot Capable of Transient Manoeuvres”
Unlike wheeled robots which have dominated the robotic world in the past, legged robots offer more versatility and the ability to traverse different terrains. The implementation of high speed, dynamic maneuvers such as running, jumping and turning allow legged robot the ability to maximize their mobility without sacrificing speed and stability, the same way that these maneuvers allow humans to adapt to their surroundings. This research focuses on the design, development and validation of a bipedal robot capable of performing transient maneuvers through trajectory optimization techniques.

Robotics Research Group Meeting 10 Our two speakers are:Devlon Erasmus – “Trajectory Optimisation on a Bipedal Robot”Legged robotics has become more prominent in recent years, led by the devel...

Ninth Robotics Research Group Meeting with our two speakers:
Mohamed Zahier Parker – “Automated Landing of a Fixed-Wing Aircraft onto a Moving Platform”
The recent rise of UAVs in both the civilian and military sectors has created a demand for a safe and robust autonomous landing system. This demand is created from the fact that the landing phase is statically regarded as the most dangerous stage of flight. This project aims to solve this problem by designing, simulating, implementing, and practically testing guidance and flight control systems that can navigate the aircraft on a trajectory and land it on a moving platform.

Sheldon Erasmus – “Motion Planning and Feedback Control of a Hexapod Robot”
In the past wheeled robots have been most common due to the simplicity of their design and their control systems, but they perform poorly on the uneven terrain which is common in mining, agriculture, and search and rescue environments. This thesis will aim to develop a legged hexapod robot that will be able to autonomously navigate those uneven terrains despite parameter uncertainty and external disturbances.

Robotics Research Group Meeting 9 Our two speakers are:Mohamed Zahier Parker – “Automated Landing of a Fixed-Wing Aircraft onto a Moving Platform”The recent rise of Unmanned Aerial Vehicles (...