Relevant Degree Programs
Affiliations to Research Centres, Institutes & Clusters
Ian M. Mitchell received a B.A.Sc. in Engineering Physics and an M.Sc. in Computer Science from the University of British Columbia, Canada in 1994 and 1997 respectively, and a Ph.D. in Scientific Computing and Computational Mathematics from Stanford University in 2002. After spending a year as a postdoctoral researcher in the Department of Electrical Engineering and Computer Science at the University of California, Berkeley and the Department of Computer Science at Stanford, Dr. Mitchell joined the faculty in the Department of Computer Science at the University of British Columbia where he is now a professor. He is the author of the Toolbox of Level Set Methods, the first publicly available high accuracy implementation of solvers for dynamic implicit surfaces and the time dependent Hamilton-Jacobi equation that works in arbitrary dimension. His research interests include development of algorithms and software for nonlinear differential equations, formal verification, control and planning in cyber-physical and robotic systems, assistive technology, and reproducible research.
Complete these steps before you reach out to a faculty member!
- Familiarize yourself with program requirements. You want to learn as much as possible from the information available to you before you reach out to a faculty member. Be sure to visit the graduate degree program listing and program-specific websites.
- Check whether the program requires you to seek commitment from a supervisor prior to submitting an application. For some programs this is an essential step while others match successful applicants with faculty members within the first year of study. This is either indicated in the program profile under "Admission Information & Requirements" - "Prepare Application" - "Supervision" or on the program website.
- Identify specific faculty members who are conducting research in your specific area of interest.
- Establish that your research interests align with the faculty member’s research interests.
- Read up on the faculty members in the program and the research being conducted in the department.
- Familiarize yourself with their work, read their recent publications and past theses/dissertations that they supervised. Be certain that their research is indeed what you are hoping to study.
- Compose an error-free and grammatically correct email addressed to your specifically targeted faculty member, and remember to use their correct titles.
- Do not send non-specific, mass emails to everyone in the department hoping for a match.
- Address the faculty members by name. Your contact should be genuine rather than generic.
- Include a brief outline of your academic background, why you are interested in working with the faculty member, and what experience you could bring to the department. The supervision enquiry form guides you with targeted questions. Ensure to craft compelling answers to these questions.
- Highlight your achievements and why you are a top student. Faculty members receive dozens of requests from prospective students and you may have less than 30 seconds to pique someone’s interest.
- Demonstrate that you are familiar with their research:
- Convey the specific ways you are a good fit for the program.
- Convey the specific ways the program/lab/faculty member is a good fit for the research you are interested in/already conducting.
- Be enthusiastic, but don’t overdo it.
G+PS regularly provides virtual sessions that focus on admission requirements and procedures and tips how to improve your application.
Graduate Student Supervision
Doctoral Student Supervision
Dissertations completed in 2010 or later are listed below. Please note that there is a 6-12 month delay to add the latest dissertations.
Reachability analysis and viability theory are key in providing guarantees of safety and proving the existence of safety-preserving controllers for constrained dynamical systems. The minimal reachable tube and (by duality) the viability kernel are the only constructs that can be used for this purpose. Unfortunately, current numerical schemes that compute these constructs suffer from a complexity that is exponential in the dimension of the state, rendering them impractical for systems of dimension greater than three or four.In this thesis we propose two separate approaches that improve the scalability of the computation of the minimal reachable tube and the viability kernel for high-dimensional systems. The first approach is based on structure decomposition and aims to facilitate the use of computationally intensive yet versatile and powerful tools for higher-dimensional linear time-invariant (LTI) systems. Within the structure decomposition framework we present two techniques – Schur-based and Riccati-based decompositions – that impose an appropriate structure on the system which is then exploited for the computation of our desired constructs in lower-dimensional subspaces.The second approach is based on set-theoretic methods and draws a new connection between the viability kernel and maximal reachable sets. Existing tools that compute the maximal reachable sets are efficient and scalable with polynomial complexity in time and space. As such, these scalable techniques can now be used to compute our desired constructs and therefore provide guarantees of safety for high-dimensional systems. Based on this new connection between the viability kernel and maximal reachable sets we propose a scalable algorithm using ellipsoidal techniques for reachability. We show that this algorithm can efficiently compute a conservative under-approximation of the viability kernel (or the discriminating kernel when uncertainties are present) for LTI systems. We then propose a permissive state-feedback control strategy that is capable of preserving safety despite bounded input authority and possibly unknown disturbances or model uncertainties for high-dimensional systems.We demonstrate the results of both of our approaches on a number of practical examples including a problem of safety in control of anesthesia and a problem of aerodynamic flight envelope protection.
The solution of a static Hamilton-Jacobi Partial Differential Equation (HJ PDE) can be used to determine the change of shape in a surface for etching/deposition/lithography applications, to provide the first-arrival time of a wavefront emanating from a source for seismic applications, or to compute the minimal-time trajectory of a robot trying to reach a goal. HJ PDEs are nonlinear so theory and methods for solving linear PDEs do not directly apply. An efficient way to approximate the solution is to emulate the causal property of this class of HJ PDE: the solution at a particularpoint only depends on values backwards along the characteristic that passes through that point and solution values always increase along characteristics. In our discretization of the HJ PDE we enforce an analogous causal property, that the solution value at a grid node may only depend on the values of nodes in its numerical stencil which are smaller. This causal property is related but not the same thing as an upwinding property of schemes for time dependent problems. The solution to such a discretized system of equations can be efficiently computed using a Dijkstra-like method in a single pass through the grid nodes in order of nondecreasing value. We develop two Dijkstra-like methods for solving two subclasses of static HJ PDEs. The first method is an extension of the Fast Marching Method for isotropic Eikonal equations and it can be used to solve a class of axis-aligned anisotropic HJ PDEs on an orthogonal grid. The second method solves general convex static HJ PDEs on simplicial grids by computing stencils for a causal discretization in an initial pass through the grid nodes, and then solving the discretization in a second Dijkstra-like pass through the nodes. This method is suitable for computing solutions on highly nonuniform grids, which may be useful for extending it to an error-control method based on adaptive grid refinement.
Master's Student Supervision
Theses completed in 2010 or later are listed below. Please note that there is a 6-12 month delay to add the latest theses.
Allocating authority appropriately between humans and machines in shared control applications is crucial for the performance of the system. Particularly in the context of collaborative wheelchairs, the arbitration should be sensitive to user needs and preferences in order to avoid confusion and frustration. Current approaches to shared control for wheelchair navigation have been designed to handle objective and functional information such as goals and system states with limited analyses to subjective information such as the user’s feelings when an assisted driving intervention is introduced. This thesis explores user affective responses on smart-wheelchairs as a potential communication channel through which users could interact more effectively with their smart mobility device. We present an implementation of shared control paradigms from the smart-wheelchair literature and results from a study where participants reported their affective interpretation of the emerging behaviours.
An autonomous or semi-autonomous powered wheelchair would bring the benefits of increased mobility and independence to a large population of cognitively impaired older adults who are not currently able to operate traditional powered wheelchairs. Algorithms for navigation of such wheelchairs are particularly challenging due to the unstructured, dynamic environments older adults navigate in their daily lives. Another set of challenges is found in the strict requirements for safety and comfort of such platforms. We aim to address the requirements of safe, smooth, and fast control with a version of the gradient sampling optimization algorithm of [Burke, Lewis & Overton, 2005]. We suggest that the uncertainty arising from such complex environments be tracked using a particle filter, and we propose the Gradient Sampling with Particle Filter (GSPF) algorithm, which uses the particles as the locations in which to sample the gradient. At each step, the GSPF efficiently finds a consensus direction suitable for all particles or identifies the type of stationary point on which it is stuck. If the stationary point is a minimum, the system has reached its goal (to within the limits of the state uncertainty) and the algorithm naturally terminates; otherwise, we propose two approaches to find a suitable descent direction. We illustrate the effectiveness of the GSPF on several examples with a holonomic robot, using the Robot Operating System (ROS) and Gazebo robot simulation environment, and also briefly demonstrate its extension to use a version of the RRT* planner instead of a value function.
Mobility is one of the most significant factors that determines older adults’ perceived level of health and well being. Cognitively impaired older adults are deprived of using powered wheelchairs because of the operational safety risks. These users can benefit from intelligent assistance during cognitively or visually challenging tasks such as back-in parking. An intelligent powered wheelchair that assists a cognitively impaired elderly user to perform a back-in parking task is proposed. A single subject participatory action design method is used with a cognitively impaired older adult to identify design guidelines for the proposed system. Based on analysis of transcripts from semi-structured interviews with the participant, a semi-autonomous back-in parking system is designed to drive the powered wheelchair into a pre-specified back-in parking space when the user commands it to. A prototype of a non-intrusive steering guidance feature for a joystick handle is also designed to render shear force in a way that can be associated with steering behavior of a car. The performance of the proposed system is evaluated in a pilot study. Experiments with the autonomous trigger and autonomous assisted modes are conducted during a back-in parking task with real-life obstacles such as tables and chairs in a long-term care facility. A single-subject research design is used to acquire and analyze quantitative data as a pilot study. Results demonstrate an increase in the user’s perception of ease of use, effectiveness and feeling of safety with the proposed system. While the user experienced at least one minor contact in 37.5% of the trials when driving unaided, the proposed system eliminated all minor contacts. No statistically significant difference in completion time and route length is observed with the proposed system. In the future, improved back-in parking systems can use this work as a benchmark for single subject participatory action design. Future iterations could also replicate the usability study on a larger population.
Smart powered wheelchairs offer the possibility of enhanced mobility to a large and growing population---most notably older adults---and a key feature of such a chair is collision avoidance. Sensors are required to detect nearby obstacles; however, complete sensor coverage of the immediate neighbourhood is challenging for reasons including financial, computational, aesthetic, user identity and sensor reliability. It is also desirable to predict the future motion of the wheelchair based on potential input signals; however, direct modeling and control of commercial wheelchairs is not possible because of proprietary internals and interfaces. In this thesis we design a dynamic egocentric occupancy map which maintains information about local obstacles even when they are outside the field of view of the sensor system, and we construct a neural network model of the mapping between joystick inputs and wheelchair motion. Using this map and model infrastructure, we can evaluate a variety of risk assessment metrics for collaborative control of a smart wheelchair. One such metric is demonstrated on a wheelchair with a single RGB-D camera in a doorway traversal scenario where the near edge of the doorframe is no longer visible to the camera as the chair makes its turn.
Daily patterns of behaviour are a rich source of information and play animportant role in establishing a person’s quality of life. Lifespace refers tomeasurements of the frequency, geographic extent and independence of anindividual’s travels. While difficult to measure and record automatically,lifespace has been shown to correlate to important metrics relating to physical performance, nutritional risk, and community engagement.MobiSense is a mobile health research platform that aims to improve mobility analysis for both ambulating and wheelchair users. The goals of thesystem were to be simple for users to collect mobility data, provide accessiblesummaries of daily behaviours and to enable further research and development in this area. The system is capable of lifespace summaries relating toindoor and outdoor mobility as well as activity trends and behaviours.For indoor reporting, we investigated robust classification algorithms forroom level indoor localization using WiFi signal strengths. We pursue topological map localization as it requires simpler map models while preservinguseful semantic information associated with location. Personalized mapsare easy to create by capturing training observations in areas of interest.Outdoor summaries are captured by periodically recording GPS fixes.For activity monitoring, a decision tree classifier was learned using acombination of accelerometer and GPS features. The classifier can differentiate between stationary, wheeling (in a wheelchair), walking or vehiclemotion.To capture the relevant sensor data, we extended an open source logging application which records data streams locally before uploading datato a web service to process and visualize results. The custom web serviceprocesses the data and generates summary files which can then be visualized either for each individual day or over a user selected date range. Weemployed a heat map visualization for outdoor lifespace to understand thegeographic extent of a user’s mobility. For indoor and activity summaries,we employed temporal line charts to understand trends in a user’s mobility.