Filter results

47688 results

ELLIS Delft Talk by Javier Alonso-Mora

ELLIS Delft Talk by Javier Alonso-Mora 12 April 2022 16:00 (NOTE: the meeting moved from 5th -> 12th) This will be a hybrid meeting This meeting is open for all interested researchers. Motion Planning among Decision-Making Agents: Trajectory Optimization with Learned Cost Functions Abstract We move towards an era of smart cities, where autonomous vehicles will provide on-demand transportation while making our streets safer and mobile robots will coexist with humans. The motion plan of mobile robots and autonomous vehicles must, therefore, account for the interaction with other agents and consider that they are, as well, decision-making entities. For example, when humans drive a car, they are fully aware of their environment and how other drivers and pedestrians may react to their future actions. Towards this objective I will discuss several methods for motion planning and multi-robot coordination that leverage constrained optimization and reinforcement learning to achieve interactive behaviors with safety guarantees. Namely: using inverse reinforcement learning and social value estimation to achieve social behaviors; employing a learned policy to guide the motion planner in dense traffic scenarios or for information gathering; achieving social trajectories by learning a cost function from a dataset of human-driven vehicles; and learning to communicate the relevant information for multi-robot coordination. The methods are of broad applicability, including autonomous vehicles and aerial vehicles. Bio Javier Alonso-Mora is an Associate Professor at the Department of Cognitive Robotics of the Delft University of Technology, the director of the Autonomous Multi-robots Laboratory, a Principal Investigator at the Amsterdam Institute for Advanced Metropolitan Solutions and co-founder of The Routing Company. Previously, he was a Postdoctoral Associate at the Computer Science and Artificial Intelligence Lab (CSAIL) of the Massachusetts Institute of Technology. He received his Ph.D. degree in robotics from ETH Zurich, in partnership with Disney Research Zurich. He serves as associate editor for Springer Autonomous Robots, and has served as associate editor for the IEEE Robotics and Automation Letters, the Publications Chair for the IEEE International Symposium on Multi-Robot and Multi-Agent Systems 2021 and associate editor for ICRA, IROS and ICUAS. He is the recipient of several prizes and grants, including an ERC Starting Grant (2021), the ICRA Best Paper Award on Multi-robot Systems (2019), an Amazon Research Award (2019) and a talent scheme VENI award from the Netherlands Organisation for Scientific Research (2017). More info: https://www.autonomousrobots.nl/ To join this event, please contact Frans Oliehoek .

van Duijvenbode, J.R.

Profile TU Delft (2018 – current) Ph.D. candidate in Resource Engineering I obtained a MSc degree in the European Mining Course (EMC) from Delft University of Technology, Aalto University and RWTH Aachen. My master thesis was about: Development and Validation of Short-term Mine Planning Optimization Algorithms for a Sublevel Stoping Operation with Backfilling. Research PhD research into the behavioural Geology – Understanding how differences in geology influence metallurgical performance. The research topic consists of integrating collected information on metallurgical properties, directly or through proxies back into the resource model. The consideration of metallurgical costs is the only way forward to obtain truly optimized mining decisions, accounting for constraints and bottlenecks in the comminution circuit and chemical processing plant. This is important to better characterize metallurgical behavior of the plant feed, which allows for a morel optimal selection of process control settings. The envisioned solution will result in an increased recovery in combination with a lower utilization of energy and chemicals per tonne of processed material (lower environmental footprint). Consequently, overall OPEX will drop making lower grade ore economic while increasing the mineral resources that are available for conversion to ore reserves (lesser need to open up new mines). Moreover, a better characterization of mining blocks reduces the unintended processing of waste due to lower overall classification errors. Copromotor: Dr. M. Soleymani Shishvan Promotor(s): Dr. M. Buxton and Prof. Jan Dirk Jansen Jeroen van Duijvenbode PhD Candidate + 31 15 27 82262 J.R.vanDuijvenbode@tudelft.nl Faculty of Civil Engineering and Geosciences Building 23 Stevinweg 1 / PO-box 5048 2628 CN Delft / 2600 GA Delft Room number: 3.21

ELLIS Delft Talk by Guillaume Rongier

ELLIS Delft Talk by Guillaume Rongier Going beyond empirical relationships in geology: The example of total organic carbon 01 February 2022 16:00 Abstract While machine learning has a long history in geology, empirical relationships remain widely used. Through the example of total organic carbon (TOC), this talk will illustrate the close links between empirical relationships and machine learning, and the benefits of turning to machine learning. TOC is a measure of the proportion of organic carbon in rock samples typically gathered from boreholes. It can be used to assess the potential for hydrocarbons, understand rock mechanics, or assess reducing conditions for basin-hosted mineral systems, and is paramount when seeking to understand variations in paleo-environmental conditions. Since gathering and analyzing rock samples is expensive, empirical relationships have been developed to predict TOC from well logs, which are based on more widely available geophysical measurements into boreholes. Those empirical relationships come from geological and petrophysical principles implemented in mathematical models manually fitted to the data. This leads to several limitations, mainly poor generalization, inability to quantify uncertainties, time-consuming and subjective calibration that leads to reproducibility issues. But those empirical relationships can be rewritten as linear regressions, a simple change that solves many of the previous limitations. Turning to more advanced machine learning methods improves predictions by taking into account the non-linearity and variability in the data. Using the expert knowledge behind empirical relationships as input besides well logs improves the predictions as well: this shows that leveraging geological and petrophysical concepts through feature selection and engineering boosts machine learning performances. To join this event, please contact Frans Oliehoek .

Half Height Horizontal

New LDE trainee in D&I office

Keehan Akbari has started since the beginning of September as a new LDE trainee in the Diversity and Inclusion office. What motivated him to work for the D&I office, what does he expect to achieve during this traineeship? Read the short interview below! What motivated you to pursue your LDE traineeship in Diversity and Inclusion office of the TU Delft? I completed both bachelor's and master's degrees in Cultural Anthropology and Development Sociology at Leiden University. Within these studies, my main area of interest was in themes of inclusion and diversity. After being hired as a trainee for the LDE traineeship, and discovering that one of the possible assignments belonged to the Diversity and Inclusion office, my choice was quickly made. I saw this as an excellent opportunity to put the theories I learned during my studies into practice. What specific skills or experiences do you bring to the D&I office that will help promote inclusivity on campus? I am someone who likes to connect rather than polarize, taking into account the importance of different perspectives and stakeholders. I believe that this is how one can achieve the most in fostering diversity and inclusion. You need to get multiple parties on board to get the best results. What are your main goals as you begin your role here, and how do you hope to make an impact? An important goal for me this year is to get students more involved in diversity and inclusion at the university. One way I will try to accomplish this is by contributing to the creation of D&I student teams. By establishing a D&I student team for faculties, it will be possible to deal with diversity- and inclusion-related issues that apply and relate to the specific department. How do you plan to engage with different (student) communities within the university? Since I am new to TU Delft, the first thing I need to do is expand my network here. Therefore, I am currently busy exploring the university and getting to know various stakeholders. Moreover, I intend to be in close contact with various student and study organizations to explore together how to strengthen cooperation on diversity and inclusion. Welcome to the team Keehan and we wish you lots of success with your traineeship!

Researchers from TU Delft and Cambridge University collaborate on innovative methods to combat Climate Change

For over a year and a half, researchers from TU Delft and the Cambridge University Centre for Climate Repair have worked together on groundbreaking techniques to increase the reflectivity of clouds in the fight against global warming. During a two-day meeting, the teams are discussing their progress. Researchers at Cambridge are focusing on the technical development of a system that can spray seawater, releasing tiny salt crystals into the atmosphere to brighten the clouds. The team from TU Delft, led by Prof. Dr. Ir. Herman Russchenberg, scientific director of the TU Delft Climate Action Program and professor of Atmospheric Remote Sensing, is studying the physical effects of this technique. Prof. Russchenberg emphasizes the importance of this research: "We have now taken the first steps towards developing emergency measures against climate change. If it proves necessary, we must be prepared to implement these techniques. Ideally, we wouldn't need to use them, but it's important to investigate how they work now." Prof. Dr. Ir. Stefan Aarninkhof, dean of the Faculty of Civil Engineering and Geosciences, expresses pride in the team as the first results of this unique collaboration are becoming visible. If the researchers in Delft and Cambridge can demonstrate the potential of the concept, the first small-scale experiments will responsibly begin within a year. This research has been made possible thanks to the long-term support from the Refreeze the Arctic Foundation, founded by family of TU Delft alumnus Marc Salzer Levi . Such generous contributions enable innovative and high-impact research that addresses urgent global challenges like climate change. Large donations like these enable the pursuit of innovative, high-impact research that may not otherwise be feasible, demonstrating how our collective effort and investment in science can lead to real, transformative solutions for global challenges like climate change. Climate-Action Programme

How system safety can make Machine Learning systems safer in the public sector

Machine Learning (ML), a form of AI where patterns are discovered in large amounts of data, can be very useful. It is increasingly used, for example, in chatbot Chat GPT, facial recognition, or speech software. However, there are also concerns about the use of ML systems in the public sector. How do you prevent the system from, for example, discriminating or making large-scale mistakes with negative effects on citizens? Scientists at TU Delft, including Jeroen Delfos, investigated how lessons from system safety can contribute to making ML systems safer in the public sector. “Policymakers are busy devising measures to counter the negative effects of ML. Our research shows that they can rely much more on existing concepts and theories that have already proven their value in other sectors,” says Jeroen Delfos. Jeroen Delfos Learning from other sectors In their research, the scientists used concepts from system safety and systems theory to describe the challenges of using ML systems in the public sector. Delfos: “Concepts and tools from the system safety literature are already widely used to support safety in sectors such as aviation, for example by analysing accidents with system safety methods. However, this is not yet common practice in the field of AI and ML. By applying a system-theoretical perspective, we view safety not only as a result of how the technology works, but as the result of a complex set of technical, social, and organisational factors.” The researchers interviewed professionals from the public sector to see which factors are recognized and which are still underexposed. Bias There is room for improvement to make ML systems in the public sector safer. For example, bias in data is still often seen as a technical problem, while the origin of that bias may lie far outside the technical system. Delfos: “Consider, for instance, the registration of crime. In neighbourhoods where the police patrol more frequently, logically, more crime is recorded, which leads to these areas being overrepresented in crime statistics. An ML system trained to discover patterns in these statistics will replicate or even reinforce this bias. However, the problem lies in the method of recording, not in the ML system itself.” Reducing risks According to the researchers, policymakers and civil servants involved in the development of ML systems would do well to incorporate system safety concepts. For example, it is advisable to identify in advance what kinds of accidents one wants to prevent when designing an ML system. Another lesson from system safety, for instance in aviation, is that systems tend to become more risky over time in practice, because safety becomes subordinate to efficiency as long as no accidents occur. “It is therefore important that safety remains a recurring topic in evaluations and that safety requirements are enforced,” says Delfos. Read the research paper .