Transforming Technologies (Past Projects)

2022 projects

  • SPAC prediction utilizing Twitter stance

    SPACs are not new to the stock market, but with the soaring market over the past two years, an increasing number of public investors and institutions are embracing the trend. This is because they could purchase SPACs at a relatively low-price point comparing stocks that going public in traditional IPO. The goal of this project is to predict the SPAC price based on the tweets stance of the merger news. We present a system which could incorporate tweet stance and historical stock prices to predict its future stock price. Through extensive experiments and detailed result analysis that our stance detection system benefits from financial information and provide some insight for public investors on SPAC stocks.

    Watch video 

    Group member:

    • Daolin Chen
  • Drug efficacy Machine Learning Model

    Response to cancer treatment is a complex phenomenon that depends on several factors. Drug susceptibility and efficacy assessment are critical for successful treatment of cancer patients. Although several clinical factors associated with drug response are now being assessed, clinical trials are time-consuming and extremely expensive.

    Develop machine learning models for drug efficacy and sensitivity prediction. Models can be used as preclinical models to understand drug response and save time, money and even lives. The required data will be provided by Platform AI.

    Watch video 

    Group member:

    • Juncheng Yu
  • Adaptation in Microservices

    Microservices is a collection of small-scale services, each of which runs its own process and communicates via a simple lightweight mechanism (usually an HTTP API). Microservices can be used for efficiently scaling computing resources in cloud systems. The main aim of this project is to empower Microservices to evolve at runtime according to user demands on a large scale. We are developing an architecture based on a MAPE-K loop that self-adapts according to the adaptation goals and performs the corresponding adaptation tasks.

    Adaptation Goals refer to the ideal settings of the system that the self-adaptation ecosystem tries to achieve, by capturing the desired state in metrics specific to the prototype implementation. To demonstrate the implementation of large-scale adaptation of microservices we will be using a ‘Video management service’ which will be developed as a case study to help us observe the metrics and adaptation goals.

    Watch video 

    Group member:

    • Prashant Singh
  • Automating Articulation Work in SPM

    Ever wanted to automate tedious tasks?

    The threats of not updating software, especially operating systems like Windows Servers is well documented. Large organizations can have a lot of systems to update and not breaking existing subsystems and apps while updating correctly without errors. So, Software Security Patch Management processes can take a lot of time. Software security patch management is a multifaceted process of identifying, acquiring, testing, installing, and verifying security patches for software products and systems. So, we wanted to automate one particular process which takes a lot of time - acquiring patches, particularly focusing on identifying the pre-requisites in the patches and coordinating patch scheduling and rescheduling tasks. After talking to a few industry experts, researching a few different scheduling algorithms and APIs from Google and Microsoft, we were able to design and implement a system that can provide automation support for articulation work in Security Patch management. 

    Watch video 

    Group members

    • Balraj Chatha
    • Thiem Van Tran Dinh
    • Prateek Arora
  • Extraction of financial events

    The project aims to use machine learning techniques to enable the extraction of events from financial documents, which help predict the future tendency of the stock market or provide investment advice. In reality, the stock market fluctuates, as its influencing factors are numerous and complicated, such as company financial reports, national policies, influential shareholders, current news, and professional inferences. The project uses the Bert technology to recognize named entities. The adopted method is BIO, where B (begin) represents the beginning of the entity, I (inside) means the internal part of the entity, and O (outside) represents the non-entity part. The entities of the project include 48 roles in total. For example, the sentence is "Securities Code: 300142 Securities Abbreviation: Watson Biological Bulletin No.: 2016-072". The entity '300142' with the role of 'StockCode' and the entity 'Watson Biological' with the role of 'StockAbbr' are obtained.

    Watch video 

    Group member

    • Qingying Zhao
  • DevOps AI/ML models for Robots

    This project aims at developing DevOps based platform that can be used for training AI/ML models using the continuous stream of input data and deploying the trained models on the target robots or edge devices. The students will be working on images and other data sets related to the operating environment. The data can be collected from a number of different input sources including cameras, still images, sensors (such as LiDAR) and descriptions of the objects. For experimentation purposes, the students can use a public or private cloud as a hosting platform and TurtleBots with Robotic Operating System as client systems. Multiple AI/ML algorithms and frameworks would be explored during the project. 

    A working prototype developed as an outcome of the research can be hosted on a public or a private cloud and can be used to train the given models with new datasets within given time constraints.

    Watch video 

    Group members:

    • Borna Morassaei
    • Lefei Mei
    • Xiangyu Shi
  • Fake news detection

    Misinformation misleads one’s opinion formation, which adversely affects and influences the decision making procedure in numerou domains (e.g., in economic, health, election
    scenes). Manual misinformation checking is a tedious and time-consuming task. However, machine learning (ML) techniques, especially fueled by recent advances in deep learning, have gained great attention and shown promising performance in misinformation detection.

    Building upon ML techniques, this work examines a different angle to detection of misinformation through truth discovery. Due to the huge amount of misinformation on the Internet, manual labeling of misinformation has become an impossible task. Therefore, we propose a semi-supervised learning method. First, the semi-supervised learning method is used to find the truth value of the data, and then, the misinformation is judged by calculating the cosine similarity between the data and the ground truth.

    Watch video 

    Group member:

    • Weiqin Yang
  • Context-based adaptive tech for MS

    In nowadays, microservice architectures have become the most popular architecture in industrial practice. However, because microservices are deployed in clusters, each service could correlate to multiple microservice instances, which leaves difficulties in responses and invocations. This project researches the microservice invocation technology from the adjustment of the service request queue and the selection of microservice instances. To improve the accuracy of microservice instance selection and the execution efficiency of microservices, and further realize high-quality microservice applications.

    Watch video 

    Group member:

    • Wenhao Zhao
  • Segment an image by limited data

    Semantic segmentation could be an essential pre-process for some tasks, such as medical diagnostics and object detection. It finishes the computer vision task by assigning each pixel a label with a corresponding class. 

    However, when the image has a high resolution, the semantic segmentation process becomes expensive and time-consuming. Therefore, Weakly-supervised semantic segmentation aims to use ‘less’ information to accomplish the similar task with a lower annotation cost. Different from the fully supervised setting of the semantic segmentation, weakly semantic segmentation focuses on the image-level labels instead of assigning every pixel on the image with related labels.   

    This project aims to train a semantic segmentation model by using weakly annotated image-level labels and achieves a satisfying performance. After that, we can get a segmentation model which can be an aid for Medical Image Analysis and Self-driving automobile.

    Watch video 

    Group member:

    • Xiaotong Wang
  • Large-scale exploration of DCS

    A microservice framework is a distributed complex system, and this research studies a microservice framework named Mape-k. First of all, for large-scale exploration, the system needs to have a certain adaptability to adapt to changes in various situations in a large-scale state. Therefore, the system implements CPU, memory usage, and number of connections for each microservice, and 403 error adaptation. The experiment realized the original limit of 4 microservices, and the operation of more microservices can be realized.

    Watch video 

    Group member:

    • Xinhao Huang
  • Explore new roles with GPT-3 model!

    GPT-3 is the largest language model in the world today. The project identifies the role of sentences in academic writing through GPT-3 and classifies sentences by roles (e.g., literature abstract, purpose of research ......). The aim is to explore the sentence roles that can be generated by sentences extracted by GPT-3 from academic writing and to train a prediction language model that reproduces GPT-3 predictions. Let's see how many sentence roles GPT-3 can generate when faced with academic writing!

    Watch video 

    Group member:

    • XuanLi
  • Swarm-based information sharing

    This project is dedicated to understanding the principles and functions of data ferrying. We built a model of a bee collecting nectar. Bees as agents have the ability to explore and communicate in the model. They store information after they get a message and share data when they get close to each other. This process helps us understand the role and efficacy of data ferrying. We collected simulation data under different conditions to analyze the behavior of information sharing in the model and the changes in performance. Finally, we will propose an algorithm suitable for this model, and try to summarize the suitable application environment for data ferry and why.

    Watch video 

    Group member:

    • Yunlong Lyu
  • Uncertainty for Object Detection DNN

    In this project, various uncertainty quantification methods will be implemented and compared to study whether the new methods proposed recently, e.g. evidential-theory-based method, have better effects than traditional methods such as Mont Carlo-Dropout and Dense Ensemble.

    Watch video 

    Group member:

    • Yurun Feng
  • Zero Energy Device for Positioning

    Since Global Positioning System (GPS) signals cannot be received indoors, many indoor positioning systems (IPSs) have been developed. However, most IPSs require putting a device with batteries on the objects. Thus, this project is to establish a prototype of IPS which can achieve centimetric-accuracy and simultaneously tracking of multiple battery-less objects. The focus of this project is to track specially designed tags using cameras. The principle is to make tags with fiducial markers on a retroreflective surface and the tags are sticked on objects being tracked. Then the tags can be distinguished from the unrelated objects in the background. Both the coordinates and the orientations of tags can be evaluated relative to the camera's coordinate system and the maximum object distance is 7 meters. Our prototype can now simultaneously track 586 objects moving at high speed within the range of 7 meters and the average position error is below 6 cm.

    Watch video 

    Group members:

    • Congyu Song                        
    • Wei Huang
  • Wideband Gap Three Phase Inverter

    The immediate objective of this project is to delineate the studies of innovations concerning the development of a DC-AC inverter with the goal of driving the existing 3-phase AC motor used by the AUMT in their electric vehicle motorsport application. The AUMT expressed their need to explore a new option for a DC-AC inverter due to their current inverter generating some limitations and issues. The project will uncover the fundamentals of DC-AC inverters and how to drive 3-phase AC motors. Exploration of wideband gap materials will be undertaken to allow the AUMT to drive motors that require higher current, voltage, and other variables due to their superior qualities whilst maintaining a smaller form factor compared to conventional semiconductors. Wideband gap materials use in inverters and the increased functionality will be conveyed through the use of schematics and simulations to allow stakeholders to conceptualise the benefits. 

    Watch video 

    Group members:

    • Costa Manavis
    • Guangyao Zhu
  • Ultrashort Pulsed Laser Ablation

    The aim of this project is to design and construct an ultrashort pulsed laser capable of penetrating soft tissue. Biomedical applications such as tissue ablation in precision surgery and photodynamic therapy for cancer treatment will be the focus for this project. These are two of many potential applications for ultrashort pulsed lasers and highlight their importance in scientific development. 

    From the research conducted a range of wavelength to be tested has been determined to be 400m to 2000mm, where the laser will have a pulse duration of 100fs. Ablation will be tested on non-biological materials such as silicon wafers, as well as biological materials such as chicken breast and human cornea.

    Once the ablation process has presented results, the laser will be reconfigured to allow for testing in the field of photodynamic therapy, where the goal is to activate the chemotherapy related drug in only one area of the body.

    Watch video 

    Group members:

    • Preeta Bhindi
    • Evan Saloniklis
  • Can autonomous drones replace bees?

    With global food demand increasing, greenhouses are filling the gap but required manual pollination of crops. Flapping wing drones have the potential to efficiently pollinate crops supported by other drones. The first aim of this project was to develop a flight controller for a flapping wing micro air vehicle. This second aim was to fly the flapping wing drone in consensus with a quadcopter to accomplish a common task. A combination of hardware and software was assembled to produce a flapping wing platform. Simulation was used to develop, test and tune a flight controller algorithm. Communication and control testing was then conducted with the goal of achieving flight scenarios. We present a prototype flapping wing drone capable of autonomous flight control. The consensus flight scenarios are achieved in simulation, and in practice whilst using a quadcopter in place of the flapping wing drone. 

    Watch video 

    Group members:

    • Christopher Gordon Balnaves
    • Harry Flynn
    • Matthew Wallage
  • Are two surveyors better than one?

    On paper, a simple solution to increase the speed of autonomous mapping is to use multiple mapping robots instead of one. Using two robots could theoretically half the time required to map an area, however, starting positions and the method used to merge the robots’ maps can significantly affect the mapping speed. We aim to find how these variables affect mapping speed and quality. These results can be used to determine if the performance benefits of two robots outweighs the cost. We began by using a TurtleBot3 with a LiDAR scanner to create a 2-D map of an area. A second TurtleBot3 was then added, combining its map with the original, first using the robots’ initial positions, and then using a machine learning algorithm. We collected data on the different map merging techniques and how they performed with different robot starting positions.

    Watch video 

    Group members:

    • Patrick Hickman
    • Joshua Cooke
  • 3D Printed Antennas

    In this project, ‘3D printed Dielectric Antennas for Novel Applications’, our group used additive manufacturing technologies, also known as 3D printing, along with state-of-the-art high-permittivity dielectric filaments and electromagnetic simulation software to design antennas for novel applications. 3D printing allows for much more complex shapes to be designed and simulated, and thus increases the degrees of freedom available for antenna design and allows for the rapid prototyping of antennas, speeding up the development process and significantly reducing its costs. This project involved researching the literature and producing a design for a 3D printable dielectric antenna of a specified permittivity, simulating that design in CST Studio Suite, printing the design, and then measuring its physical characteristics, such as the gain, directivity, radiation pattern and bandwidth of that antenna within the university's anechoic chamber.

    Watch video 

    Group members:

    • Shannon Sluggett
    • Christopher Blute
  • Genealogy: Finding the Somerton Man

    When trying to identify human remains, such as the Somerton Man, a visual medium is extremely useful for genealogists to track genetic histories and discover potential identities. Our job was to build a program to convert genealogical information from the industry standard GEDCOM file to an interactive map. We created the program in python to convert GEDCOM files into .csv files and then utilising several Google APIs, create an interactive, Google Earth style map. A key component of the program was the capability to reliably and reasonably exclude information that experts deemed not useful or unreliable. This included ambiguous location data, such as a city with no state, and names of individuals that appear to be placeholders rather than real names. The created map works excellently to shows the birth and death locations and dates for all individuals listed in the GEDCOM file(s), helping genealogists track down potential identities.

    Watch video 

    Group members:

    • Shaun Fernando
    • Harrison Boyce
  • Vision Based Autonomous Racing Car

    This project aims to build a ground-based vehicle that can autonomously race around a circuit, relying only on visual sensing of the circuit and environment. There is significant advancement in autonomous systems that rely on multiple sensors to be aware of the environment the system is in. Limiting the sensor suite to just visual sensors and IMUs is challenging but mimics the visual capability of a human operator.

    The development will include investigating different techniques such as Convolutional Neural Networks, Recurrent Neural Networks and Reinforcement Learning as well as building multiple vehicles that are suitable for a small-scale testing. The goal of this project is to design small-scale RC cars that can autonomously race around a circuit relying only on visual sensors and IMUs and are able to race by themselves on a track while also gaining the maximum cumulative total reward to optimise the performance.

    Watch video 

    Group members:

    • Zerui Li
    • Raghav Bhardwaj
  • Bots vs Humans Online

    The 2022 Russian invasion of Ukraine emphasises the role social media plays in modern-day conflicts, with both sides fighting across both the physical and information environments. We identify malicious cyber-activity, and consider the effect this activity has on the overall conversation on Twitter around the Russia/Ukraine Conflict. Here, we consider information flows, sentiment analysis and other linguistic analysis to understand how bot activity influences wider online discourse.

    Watch video 

    Group members:

    • Bridgett Smart
  • Fibre fabrication temperature model

    The uses of Microstructured Optical Fibres (MOFs) include communication networks, chemical sensing in hazardous environments, biological sensing of cancer and DNA detection, and temperature and pressure sensors. There are many holes inside an optical fibre. These holes are supposed to be circular. However, during the fibre fabricating process, these holes deform into an elliptical shape. My masters project aims to model the fibre fabricating process, to find the temperature needed to produce the desired fibre structure. This reduces experimental costs and time needed to obtain the desired fibre shape. Parametrisation and validation of the model I develop will be by comparison of solutions with results from experiments to be done by others. 

    Watch video 

    Group member:

    • Yunice Yuwono
  • Modelling Option Prices

    Options are a financial instrument that has had significant focus in mathematical literature. Initially, models of option prices assumed that stock prices were well approximated by a stochastic process called Brownian motion. However, several issues with these models have been discovered and the literature has been investigating models using different stochastic processes. In this project, we produce an algorithm that models the price of barrier options using a meromorphic Levy process. We achieved a high level of accuracy, with relatively fast convergence.

    Watch video 

    Group member:

    • Jesse Tonkin
  • Transgression of loops on a sphere

    Given a loop on a surface, its holonomy measures the extent to which a vector parallelly transported along the entire loop fails to agree with its initial state. This extent can be measured by a non-zero complex number, thus giving us a function which takes in a loop on a surface and returns a non-zero complex number. This is a special case of transgression: given, on a manifold M, a geometric object known as a line bundle, we obtain a function which is evaluated on loops in M. This can be generalised to higher geometric objects known as bundle gerbes, and my project is looking at generalising this further and exhibiting examples of this correspondence.

    Watch video 

    Group member:

    • Matthew Charles Hoffman
  • Digital Prototyping - SITIA Trektor

    The Digital Prototyping for SITIA Trektor project aims to implement autonomous functions in the SITIA Trektor – a robotic agricultural tractor which primarily operates in vineyards. This project aims to automate the SITIA Trektor to reduce human labour required in vineyards. The scope of autonomous function is limited to self-navigation and obstacle avoidance. The project will first develop a functional digital twin instance of the tractor in 3DS Experience using SITIA’s CAD files and live connections to the tractor’s sensors. The autonomous algorithms will be developed in an iterative process, beginning with basic obstacle avoidance along a prescribed route, and culminating in self-localization and self-navigation in an unknown, unmapped environment. The algorithms will be based on Simultaneous Localization and Mapping (SLAM) techniques, which allow a vehicle to map its immediate environment, and identify its position in the environment. 

    Watch video 

    Group member:

    • Aryan Mathur
  • Coffee Particle Characterisation

    Our endeavour is to create an affordable particle analyser for the improvement of quality and consistency within the coffee industry. The device utilises dynamic image analysis optimised for coffee characteristics to determine the size and shape of ground coffee particles. This information can help coffee shops ensure consistency of grind sizes across different grinders, diagnose grinder faults, and create more precise recipes.

    Watch video 

    Group members:

    • Dylan Warman
    • Matthew Phillips
  • From mystery to maps

    CaveX (Cave eXploration) is a multi-year engineering honours project in its second iteration, with the aim of designing and building a biologically-inspired mobile robot platform for scanning and mapping cave systems. Caves are a unique environment that present both rich scientific value and the challenge of exploring them in a safe, low-impact way. Most caves worldwide have been explored by humans, using bulky equipment that poses risks to operator safety and environment preservation. By employing advanced robotics and Light Detection and Ranging (LiDAR) technology, the CaveX robot can explore previously inaccessible cave areas, collecting new mapping data and contributing to significant research efforts. The 2022 iteration has the objective of improving the robot's usability and robustness, through the design and build of a second, refined physical prototype and implementation of new movement control software. The resulting system can more reliably transverse cave terrain, continuing to contribute valuable scientific data.

    Watch video 

    Group members:

    • Henry Bright
    • Nicholas Verboon
    • Lachlan Zilm
    • Tara Mahony
  • Evaporation technology

    Compared to traditional air conditioners, Indirect Evaporative Cooler(IEC) is more environmentally friendly. The main material of IEC—nylon polymer sheet is the core of cooler as it determines the performance of the IEC system. In this project, a testing rig is needed to design and build to help understand the effectiveness(capillary and evaporation effects) of re-surfaced nylon sheets. In this testing rig, an air-blower provides a constant speed air and passes through PVC piping to a larger air duct made of acrylic sheets. When the wind blows cross the nylon sheet, the wind will take away the moisture on the surface of sheet and reduce the temperature of the air. At this time, the water will be absorbed from the watertank by the nylon sheet and the balance can test the weight change of watertank to know the capillary and evaporation effects of nylon sheets.

    Watch video 

    Group member:

    • Jingbo Zhou
  • Can Drones Identify Weeds?

    African Boxthorn is an introduced thorny weed and is one of regional Australia’s most widespread and invasive shrubs. If left unattended, the African Boxthorn will grow to cover large areas of land, block waterways, harbour pests, injure livestock and interfere with native species. The goal of this project is to design, build and test a Boxthorn Identification Drone (BTID) system that can increase the efficiency of African Boxthorn weed removal. Both hardware and software have been used and developed to record and process Boxthorn imagery. Field days along with machine learning simulations were completed to collect data and develop a software capable of identifying the intrusive Boxthorn weed. We present a system which uses machine learning and neural networks to process imagery collected using a drone. By geo-tagging an identified Boxthorn’s location it significantly improves the efficiency when trying to locate and eradicate the Boxthorn weed.

    Watch video 

    Group members:

    • Alec Buttignol
    • Scott Bailey
    • Kale Olds
    • Mark Jenkin
  • Robotic Biodefleecing System

    This project aims to design and build an automatic sheep shearing machine that can harvest the wool from bio-treated sheep.

    Watch video 

    Group member:

    • Taige Ding
  • Single Image Object Tracking

    Imagine there was an object next you like a toy car, and you want to know the location and orientation of this object. Take one photo of the object and this could be possible. The aim of this project is to be able to track the location and orientation of an object from as few as one image of the object of interest. The tracking of the object is enabled by Neural Radiance Fields (NeRF) which are a method for representing an object allowing for the rendering of the object from new viewpoints. This rendered object from a predicted viewpoint can be compared with the actual image of the object to accurately estimate the location and orientation of the object. The results from the project allow for tracking of an object using a trained NeRF model or even from a single image of the object. 

    Watch video 

    Group member:

    • Adam Bethell
  • Machines for Smart Work

    Today, we live in a digital age where people create digital data on a daily basis, often without even recognising it. Video is a great way to get a lot of information. People in today’s world want to watch a part of that news that is relevant to their interest. Hence video retrieval of relevant news is of prime importance. In this project, we brief about various models which are in place today for video retrieval and analyse the results of a few models using news dataset. Finally based on the results we conclude which model is efficient for news video text retrieval.

    Watch video 

    Group member:

    • Puneet Valad
  • Microservice Coordination

    Defense and military operations run in a contested and dynamic environment. In such environment, the resources are very limited and the surrounding events keep changing with
    interactions. To address this contention, microservice, an architectural pattern has been used for development. This architecture improves scalability and flexibility. As the topic has gained lot of interest in the field of computer science, further research has been performed to build applications that adapt its services based on the environment. Self-adaptive microservices play a key role in overcoming the shortcomings of the traditional microservice orchestration. In our approach, we make use of a reinforcement learning algorithm that studies the behaviour of the system under various scenarios and acts accordingly in the future ensuring the efficiency of the system is not compromised.

    Watch video 

    Group member:

    • Vincy Benita Jelsingh
  • Blockchain Visualiser for Provenance

    Blockchain technology can be used to keep track of the history of digital assets, such as states and dataset. However, the current user interfaces available are not intuitive to tell user who did what to whom. Because most part of these information on the blockchain are encoded by hash and these transactions data recorded in smart contracts are hard to understand by people who has no related background. Hence, a web-based visualiser can be developed to intuitively display the relationship between those provenance data on the blockchain.

    Watch video 

    Group members:

    • Yen-Han Chen
    • Zihao Lin
    • Jack Munchenberg
    • Haorui Xu
  • Lost in space

    In a mission as important as space travel or exploration, we want to know every detail about the spacecraft as accurately as possible. This includes the direction and orientation, referred to as the attitude. There are many ways to find the attitude but most require star identification to use as a ground point. If done correctly, stars can be used to very accurately find the attitude. This gets harder when space noise causes stars to change brightness or position. In my project, I implemented the current state of the art star identification algorithm, tested it against many possible factors, and compared the results to a newly developed identification and attitude determination algorithm to see where each's strengths and weaknesses are.

    Watch video 

    Group member:

    • Ben McCarthy
  • It's that Gripper Feeling

    Human-machine interaction has been adapted to many working environments, especially in the automotive and manufacturing sectors. This is commonly operated using industrial robots that consist of a number of rigid links and hard components. Technological advancement has opened up the opportunity to approach soft robots in nowadays industry that is way safer and more reliable for working alongside humans. Therefore, we wanted to design and fabricate a soft robot gripper that possesses a high level of bio-inspiration for industry application. In this project, we used multiple conceptual designs and different shore hardness of silicone rubber in order to develop the gripper. Simulations have been conducted to assess the motion and forces of the soft robot. This project sought to provide a feasible soft robot gripper that would be actuated with a soft pneumatic system to give the desired motion for industrial purposes.

    Watch video 

    Group members:

    • Mohd Qayyum
    • Iman Sakri 
    • Muhamad Fakrul
    • Adam Mohd Muhayadin 
    • Arif Fahmi Hisam
  • Can robots learn to adapt?

    Watch video 

    Group members:

    • Jacob Drake
    • Emily Duhne
  • Attention AI

  • Big model for small data training

  • Closing Orbitisin Complex Dynamics

  • Cracking the Voynich Code

  • High Gain Omnidirectional Antenna

  • Does your Rasch Look Right

  • Electric Aircraft Acoustic Simulation

  • Its that Gripper Feeling

  • Learning without forgetting

  • Likelihood Free Bayesian Inference

  • Microservice Coordination

  • Microsystem Vibration Responses

  • Model free Prediction of Chaos

  • Post 5G Communication Technology

  • Quantam vs Classical RADAR

  • Runtime Estimator

  • Summarising Shared Information

  • Super materials of the Future

  • Terahertz Imaging in Reflection

  • The ball bearing motor mystery

  • Three wheeled solar car with monohull layout

  • We taught a computer to play games

  • Whey too much weight