2019 was our penultimate year in the Centre and one filled with recognition and accolades for our people and for our scientific achievements. Our national and international profile continues to rise, with media coverage about the Centre and our researchers reaching a global audience of well over a quarter billion people. We are leading the world in transformational research in the new field of robotic vision.
My career in robotics spans nearly 40 years and progress over that time has been astounding. We have capabilities now that we once only dreamed about, and this is finally enabling robots to perform useful tasks. But we’re not there yet, and the remaining challenges and great opportunities continue to excite me.
2019 has seen many exciting developments across the Centre in robotic capability and real-world application. Our fleet of robo reef protectors now includes the new LarvalBots which carried a large volume of coral larvae for targeted dispersal to damaged reef areas of the Great Barrier Reef. We have evolved a prototype miniaturised surgical robot called SnakeBot which can be uniquely matched to patient anatomy and could be used in future knee arthroscopy surgery. Centre researchers took an AI system on a road-trip of south-east Queensland to ensure the autonomous cars of the future will be smart enough to handle tough Australian road conditions. We’ve deployed a fleet of miniature autonomous vehicles called Carlie to create an engaging demonstrator for the public to interact with. We also launched the world’s first robotic vision challenge as an innovative way to drive global research in this area.
Using a Pepper robot, we trialled meaningful social robot interventions in healthcare settings, and this research could help with managing depression, drug and alcohol abuse, and eating disorders. We’ve seen ground-breaking research in the area of robotic grasping and manipulation which helps a robot move and think at the same time, allowing it to better see objects in a cluttered environment and therefore be more effective. The Centre competed in the RoboCup@Home league as part of RoboCup 2019 in Sydney and came sixth with our mobile domestic robot and this has driven a new research direction of mobile manipulation.
The Centre attracted over $7 million in external income this year from other Australian Research Council (ARC) grants, industry projects, collaborating organisation schemes, and other grants and awards. We had great success in the ARC Discovery Program Grant round at the end of 2019, securing $2 million in project funding for five projects across our four Australian partner institutions.
It’s been a big year media wise. We were the direct focus of more than 660 stories across print, online and broadcast media across Australia and more than 30 countries. I’d like to acknowledge and thank the efforts of our Centre Communications Specialist Shelley Thomas for her role in telling our story to the world.
As we move into our final year of operation, we have streamlined our strategic focus from five to three key strategic priorities: Science, Engage, and Culture. A critical focus for our final year is the Centre’s legacy. These are outlined below.
We are leading the world in transformational research in the new field of robotic vision.
The Centre continues to deliver breakthrough science and technologies – if the Centre were a person it would have an h-index of 77, achieved in just 6 years. We have created a new generation of robots that can visually sense and understand complex and unstructured real-world environments. We have undertaken transformational research that has achieved international standing in the fields of robotics and computer vision. Our seven research projects – Learning; Manipulation and Vision; Scene Understanding; Vision and Language; Robots, Humans and Action; Fast Visual Motion Control; and Robotic Vision, Evaluation and Benchmarking – all continue to evolve and feed into our two demonstrator projects – Self-driving Cars; and Manipulation. Of note are the influential learning algorithms that are being used for visually-guided robotic navigation systems. This includes the RefineNet algorithm from our Learning Project, which has become an important benchmark in the field and the original 2017 paper has nearly 1,000 citations. In our Fast Visual Motion Control project, the team have developed architectures and training paradigms to build state-of-the-art convolutional neural networks (CNNs) for image reconstruction and optic flow computation from event camera data streams. The Centre’s Manipulation & Vision project researchers have developed a generative grasping convolutional neural network (GG-CNN) which predicts a pixel-wise grasp quality that can be deployed in closed-loop grasping scenarios. The network has achieved excellent results in gasping, particularly in cluttered scenes, which has seen an 84% grasp success rate on a set of previously unseen objects, and 94% on household items. These are important developments, aligned with our mission, which enable robots to see, understand and act.
Our Research Projects – the teams, achievements and priorities for 2020 – are detailed in the Research Impact section of this report.
We engage with communities about the potential of robotic vision technologies by sharing our expertise and providing access to robotic vision resources.
A focus of our engagement with the scientific community has, this year, been in terms of a challenge. Used to great effect by our colleagues in the computer vision research community, we’ve thrown down the gauntlet with a truly robotic vision competition to expand the capabilities of robots that see. We challenged competitors to detect objects in video data from high-fidelity simulation of three different domestic service robots scenarios, and we compared notes and techniques at two workshops run alongside major international conferences. We will launch a followup challenge in 2020.
The Probabilistic Object Detection Challenge (PrOD) workshop was held during the Conference on Computer Vision and Pattern Recognition (CVPR) in California in June. Probabilistic object detection is important for robots to safely and effectively work in messy and unpredictable real-world environments. This was a collaboration between Centre researchers and long-standing partners from Google AI. The project team organised a second workshop on the topic of ‘The Importance of Uncertainty in Deep Learning for Robotics’ at the International Conference on Intelligent Robots and Systems (IROS) held in Macau, China in November, and have two further workshops scheduled for 2020. Competitions are very effective for driving computer vision research, but they haven’t really pushed the envelope for the type of problems that a robot may encounter. Our new competitions aim to solve that and are specifically robotic vision not computer vision challenges. This is a truly exciting development for the robotic vision community.
QUT held its fourth biennial Robotics and Technology Festival, Robotronica in August. The event attracted 22,000 people to the Gardens Point campus. I presented a talk on potential directions of robotic research and design and what that might mean for humans and society. I was really impressed by the number of thoughtful questions from young children in the audience. It’s fantastic to see this dialogue happening with our younger generation. I also got to speak to a packed tent at the Woodford Folk Festival on the topic of robots and the future of work. Finally, an exciting milestone for me was that the QUT Robot Academy turned 2 and has now delivered over 1 million lesson views.
We are creating a vibrant, energetic, future-focused and collaborative robotic vision community that is developing knowledge leaders for both industry and academia
We have worked hard since our inception to create an environment that supports our people, and enables high-quality research and innovation. Our culture encourages collaboration and communication, exploring challenges from all sides and perspectives to find transformational solutions, and developing future knowledge leaders for industry and academia. Our membership is now over 200 strong and we have an Alumni network of over 70 people.
The Centre has two annual events that are key to building our culture, namely the Robotic Vision Summer School (RVSS) and our annual symposium, RoboVis. We staged the Centre’s fifth RVSS in Kioloa in February attracting 74 attendees and 5 international guest speakers including Seth Hutchinson (Georgia Tech), José Neira Parra (Universidad de Zaragoza), Laura Leal-taixé (Technical University Of Munich), Tarek Hamel (University Of Nice Cote D’azur), and Silvère Bonnabel (Mines Paris Tech). We held our sixth annual RoboVis conference in the beautiful beachside suburb of Glenelg in Adelaide in September. The program included twelve 3MT presentations where PhD researchers present their thesis in 3 minutes, 14 demo sessions and an impressive 45 poster presentations. The Centre was extremely fortunate to have keynote speaker Dr Michelle Perugini join us at RoboVis and talk about her experience as an entrepreneur, academic and an internationally renowned expert in health, medical research, advanced analytics, and cognitive AI. Michelle’s talk was the highlight of our symposium and led to a very engaged question and answer session with our group.
We also encourage our students to take advantage of opportunities to travel to other international laboratories and global technology giants like Amazon Robotics to gain new insights and expertise in research approaches. Centre PhD researchers undertook internships at Omron in Tokyo, Japan; the Autonomous Driving Lab, of Baidu, Inc, in Beijing, China; Hikvision United States; and Skydio in California, United States.
Gender imbalances are acute in computer science and engineering, amongst academic faculty as well as in the student body which is our near-term future. This remains a huge issue facing our Centre, and one I encourage all Centre members to be part of the solution. We have set significant stretch targets of 50% female representation at our RoboVis symposium scheduled for November 2020 across delegates, speakers and session chairs, and will be striving to achieve them.
The Women in STEM Decadal Plan from the Australian Academy of Science was released in 2019 and is the output of a large and diverse expert working group. Key findings from the report include the need for STEM-trained people will continue to grow nationally; that Australia cannot meet national needs without increasing the number of women involved in STEM; that barriers exist for women at all levels of schooling and the workforce; and that responsibility for remedy must be shared by government, academia, industry, education sector and the community. Alarmingly, only 16% of Australia’s STEM-skilled workforce is female despite 47.5% of our workforce being female. Increasing the attendance and involvement of women in public events and conferences is critical to increasing female representation and is just one way we can work to shift the imbalance.
Brisbane hosted the inaugural Hopper Down Under in July. It was a celebration of diversity in technology in the Asia-Pacific, and the sister event of the Grace Hopper celebration, the world’s largest gathering of women technologists. It brought together women technologists at all levels, from undergraduate to CEO, along with members from industry leading companies, academia, and research, to build relationships, learn, and advance their careers. Hopper Down Under was an initiative of the Centre led by former Centre Chief Operating Officer Sue Keay, who travelled to Houston to meet with AnitaB.org in September 2018, inviting the organisation to establish an annual event in Australia. Hopper Down Under has already expanded to cover the Asia-Pacific region, and the 2019 program featured Chief Investigator Professor Elizabeth Croft, Associate Investigator Dana Kulić and Research Fellow Nicole Robinson. I look forward to seeing this event continue into the future.
2020 is the final year of our Centre and we have exciting developments underway to ensure the Centre lives on beyond that point. We have incredible momentum as a research entity, amazing brand recognition globally, and fantastic talent in our people. Each of our Australian partners has a promising new research entity and these offer ongoing opportunities for our researchers. The University of Adelaide launched the Australian Institute of Machine Learning (AIML) in 2018. The QUT Centre for Robotics, of which I am also director, commenced in January 2020. Monash University and the Australian National University both have considerable investment in robotics and AI.
At the time of writing this report in 2020, we are facing the uncertainty and disruption caused by the COVID-19 pandemic. “Unprecedented” is an overused term right now, but COVID-19 is certainly such an event. It is unlike anything we’ve experienced in our lifetime, and the health and economic effects of this are deeply concerning for many. This will of course impact on our operations in our final year and we have contingency plans in place to address this.
We are planning a Robotic Vision Summit event in conjunction with Price Waterhouse Coopers (PwC) to connect with industry and policy makers about robotics, vision and AI technology. This includes what it means for Australia, how we’re performing as a country in innovation and uptake, and what needs to change to maximize our benefit. We are keen to connect with representatives across all levels of government to express the importance of these technologies for the nation and invite them to our summit.
We also plan to make some Centre activities enduring, in particular the Robotic Vision Summer School (RVSS). We attracted sponsorship for the 2020 event which gives us confidence that it can be sustainable moving forward. Plans are currently underway to establish an incorporated association (Robotic Vision Australia) which will manage this event into the future and I’m excited to see this come to fruition.
The future for robotics in Australia is very bright and our robotic capability is strong. We do great research and apply it to real-world problems and this is what sets us apart from the rest of the world. An exciting development is the increasing interest in robotics across broad application areas, the commercial investment we are seeing, and the rise of robotics start-ups. I haven’t seen this in my career to date and it’s incredibly exciting to see the opportunities available for our researchers and the diverse career options available to them.
The Centre launched its fourth start-up this year, Lyro Robotics. It’s a Brisbane-based venture and an innovative spin-out from the Centre and the research undertaken for the 2017 Amazon Robotics Challenge win. The company is commercialising its world-leading robotic picking and packing technology for deployment in Australia’s warehouses, supply chains, and logistics operations. This is incredibly exciting.
I’d like to close by expressing my sincere thanks to all our Centre members for their continuing contributions to another outstanding year, and to the other Centre Executive members and Centre Advisory Board for their ongoing drive, support and advice particularly in these challenging times. There are a lot of unknowns for 2020 but I am confident that united together we will get through to the other side and continue to thrive.
Distinguished Professor Peter Corke, Centre Director