4-Year Fully Funded Studentship - Cooperative navigation through comings and goings: Mapping and localisation for open multi-agent systems

4-Year Fully Funded Studentship - Cooperative navigation through comings and goings: Mapping and localisation for open multi-agent systems

Quick Decision

Recommendation: Strong but verify fit. This is a technically distinctive UCL Electronic and Electrical Engineering (EEE) PhD on keeping cooperative navigation useful as members of a robot, vehicle or other multi-agent network join, leave, fail or lose contact. The advertised work connects decentralized localization from relative measurements and sparse GPS with neural implicit mapping, mathematical analysis and traffic-oriented empirical validation. It is a strong topic match for applicants prepared in control, estimation, robotics, applied mathematics or computer science who want to work across algorithm theory and implementation.

There is a decisive eligibility condition: the official 2027–28 catalogue identifies project 2531cs1050 as a standard project and says only Home fee-status candidates can apply. General UELA programme eligibility for some International candidates does not apply to this project; the international places are restricted to collaborative projects. UELA is also a competitive award round: about 50 studentships are expected across 133 listed projects. The application deadline is 8 January 2027 at 12:00 GMT, and UCL recommends submitting with both references by 16 December 2026. Before investing in the application, verify Home fee status and discuss the project’s intended sensor model, datasets and evaluation plan with the supervisor. (2027–28 project catalogue, UELA information, application portal)

Research Problem and Doctoral Work

What Is Unsolved and Why It Matters

The project starts from a practical weakness in many cooperative navigation methods: they assume a fixed set of agents. In real deployments, an electric vehicle may leave to charge and later rejoin, a robot may fail, or a satellite may move out of communication range. The system must continue to navigate while its membership and communication graph change. A review of open-agent systems describes this changing set of collaborators as a distinct source of uncertainty: the system must account for who is currently present as well as what each agent may do. (Project catalogue, Decision making in open agent systems)

The advertised doctoral work has three connected parts:

1. Cooperative localization in open networks: design decentralized algorithms that infer agents’ positions from relative measurements while relying on very limited GPS information.
2. Neural implicit mapping: repurpose these algorithms to represent surrounding space with a learned neural representation rather than only a conventional geometric map.
3. Theory and validation: develop and analyze the algorithms theoretically, then validate them empirically, with particular attention to traffic applications.

This is more than applying a standard SLAM package. Localization and mapping must remain useful while the team changes, and an implicit map must be learned and updated across a distributed system. The listing does not state which sensors generate the relative measurements, how membership events are represented, what neural field architecture is intended, or which traffic datasets, simulator or hardware will be used. Those are central design decisions to clarify.

What the Literature Suggests—and What It Does Not Promise

Recent work by the advertised supervisor is relevant but does not prescribe this PhD’s methods. A 2025 paper in Automatica studies a matching problem between two decoupled swarms: local control laws make agents track reference paths and reach pairwise matching without direct communication between the swarms. It contributes theory and simulation/experimental evidence for distributed multi-agent coordination, but it is not an open-network localization or mapping method. (Fedele, D’Alfonso and Chen, 2025)

A 2025 European Control Conference paper proposes distributed state estimation over dynamic communication graphs. It uses multi-hop exchange to reconstruct global linear-system states and avoid centralized redesign when network topology changes; its demonstration is a numerical example, not geometric localization. A 2024 CASE paper uses distributed path planning, Lyapunov analysis and split/merge topology to handle formation and obstacles, with simulation-based validation. A related 2024 paper covers online multi-robot coverage planning in dynamic environments. These works make the supervisor’s adjacency to changing networks and cooperative navigation visible while leaving open the advertised combination of membership changes, relative-measurement localization and learned mapping. (Dynamic-graph state estimation, 2025, reconfigurable-topology path planning, 2024, multi-robot coverage planning, 2024)

Independent mapping research shows why the second project objective is timely. MACIM distributes implicit Euclidean signed-distance-field mapping across agents that maintain local neural models and exchange model weights. Di-NeRF distributes NeRF optimization and refines relative poses jointly with the map. iS-MAP uses a multiscale hash-grid/plane representation and structural consistency losses for neural implicit RGB-D SLAM. These are useful precedents for representation, fusion and communication design; none establishes how to handle this project’s open membership or traffic setting. (MACIM, Di-NeRF, iS-MAP)

Possible lines of inquiry—not published project commitments—include how to maintain a shared coordinate frame when GPS is sparse, test observability as agents disappear, reconcile a returning agent’s local map, and compare geometric and implicit representations under communication limits. Robustness to delayed or missing messages, accuracy versus bandwidth, map consistency and compute cost are natural evaluation dimensions. The project description does not promise a particular dataset, physical robot platform, neural architecture or field deployment.

School, Programme and Field Context

UCL is the degree-awarding institution; the official catalogue lists EEE as the host department and Boli Chen as principal supervisor. UCL’s EEE MPhil/PhD page sets a general entry route of a relevant UK master’s degree or at least an upper-second-class UK bachelor’s degree (or overseas equivalent), with English language requirement Level 1. The generic full-time EEE MPhil/PhD route normally lasts three calendar years, begins with MPhil registration, and requires a transfer/upgrade examination before PhD registration. By contrast, this UELA project is advertised with four years of funding and UELA says students are registered for four years and expected to submit within the funded period. Confirm the registration route and thesis timing in the eventual offer rather than assuming the generic course duration and project funding period are interchangeable. (EEE MPhil/PhD course, UELA information)

The institution has a broad research signal: UCL reports #8 overall in the QS World University Rankings 2027 and #2 in the UK for research power in REF 2021. The EEE course page separately reports #2 in the UK for Electronic and Electrical Engineering in the Times Good University Guide 2026. These measures differ in date, method and scope; the overall rank is not a subject rank, and none measures an individual project or guarantees a particular training experience. The Faculty of Engineering also reports substantial impact ratings for its REF 2021 case studies. (UCL rankings, EEE course and subject ranking, Faculty of Engineering REF 2021)

The project sits near UCL EEE research on control, signal processing, estimation, robotics and autonomous vehicles. Chen is associated with the Sensors Systems and Circuits group and the cross-faculty UCL Robotics Institute. These links indicate an adjacent departmental ecosystem, not a confirmed co-supervision arrangement or guaranteed access to every listed facility. The EEE programme page describes a doctoral community of more than 100 PhD students and broad connections across academia and industry. UCL’s engineering AI overview adds institutional activity in machine learning and autonomous systems. (Control and Signal Processing, Sensors Systems and Circuits, UCL Robotics Institute, UCL Engineering AI research)

Supervisor and Lab Assessment

The official catalogue names Boli Chen as principal supervisor. UCL currently describes him as an Associate Professor in Electronic Engineering, with research on control, optimization and estimation of complex dynamical systems and applications in smart cities, including transport and sensor networks. The profile associates him with Sensors Systems and Circuits and the UCL Robotics Institute. His recent papers show relevant activity across multi-agent coordination, changing communication graphs and robot planning; they should be read as evidence of research adjacency, not a promise of project-specific infrastructure or a fixed thesis method. (Boli Chen’s UCL profile, official publications list)

The public pages do not specify a co-supervisor, meeting frequency, number of doctoral researchers in the immediate team, compute allocation, robot access or the project’s traffic data. The Sensors Systems and Circuits group lists general test facilities, but this does not establish that those resources are assigned to this studentship. Ask which supervisor will guide the neural mapping component, whether there is a second supervisor for robotics/SLAM, and what practical test environment is committed.

Funding, Application, and Living Cost

Award and Eligibility

This is one project within a competitive UCL EPSRC Doctoral Landscape Award round. The 2027–28 catalogue lists 133 projects—46 collaborative and 87 standard—and says approximately 50 studentships are expected, with a maximum of 15 International offers. The programme-level rules say International fee-status candidates can apply only to collaborative projects. The catalogue specifically marks 2531cs1050 standard and states “Only Home fees candidates can apply.” Therefore an applicant assessed by UCL as Overseas fee status is not eligible for this particular project, even though the UELA programme overall has some International places. UCL Graduate Admissions formally assesses fee status based on the relevant rules, including nationality, immigration status and residence; applicants should not infer status from nationality alone. (Project catalogue, UELA programme rules, UCL fee-status assessment)

If awarded, UELA provides four years of Home-rate fees, an enhanced maintenance stipend and a Research Training Support Grant for training, project costs, conferences and travel. The UCL programme page currently estimates £26,022 per year for 2027/28 in its summary, but its detailed funding paragraph labels the same £26,022 amount 2026/27 and says it is £1,500 above the £24,522 UKRI London-weighted rate. The project description repeats that 2026/27 wording. This internal year-label conflict means £26,022 should be treated as a current estimate, not a confirmed 2027/28 award rate or guaranteed amount for each funded year. Ask UELA to confirm the correct start-year figure, annual uplift schedule and any project-specific supplement in writing. The 2027/28 EEE course page currently lists the UK research fee as “to be confirmed”; the UELA award would cover the Home-rate fee if made. (UELA funding details, 2027–28 fee schedules, EEE 2027/28 course fees)

Deadline and Selection

The application portal opened on 26 October 2026 and closes on 8 January 2027 at 12:00 GMT. The portal recommends that candidates submit early and ensure both references arrive by 16 December 2026, because UCL will be closed over the Christmas period. The 2027–28 catalogue allows candidates to apply to up to three projects and requires two references by the final deadline. Selection is based on academic merit and proceeds through project shortlisting, interview and central UELA panel ranking. A project may be withdrawn before or after the application deadline. A project application is therefore an application to compete for an award, not a funding guarantee; successful candidates must also complete the formal UCL admissions process. (Application portal, UELA information and process, project catalogue)

London Budget

UCL’s indicative guide lists average postgraduate self-catered hall rent at £337 per week, or £17,524 for a 52-week year. Its listed non-rent essentials—food, course materials, mobile, health/wellbeing and a Zone 1–2 student travelcard—add about £4,904 per 52-week year, for a derived total of roughly £22,428. This is only an illustrative subtotal: UCL says the rent average is based on 2025/26 accommodation, spending estimates draw on an earlier survey uprated by 10%, rent varies by hall, and discretionary expenses are extra. UCL’s separate broad graduate living-cost guide says around £20,000, also as an estimate. On that evidence, a £26,022 headline stipend could leave only about £3,600 against the listed essentials, before discretionary costs, moving/setup expenses or 2027/28 inflation. Applicants should build an individual 12-month budget and confirm the stipend. (UCL cost guide, UCL graduate living-cost overview)

Applicant Fit and Preparation

The formal entry requirement is a relevant master's or at least a 2:1 bachelor's degree (or overseas equivalent), plus UCL English language Level 1. The catalogue does not publish a separate required robotics or neural-network qualification. It says the doctoral researcher will develop expertise in learning, open multi-agent systems, algorithm development and analysis, and traffic applications. Do not mistake the following useful preparation for official eligibility criteria: mathematical foundations (linear algebra, probability and optimization), graph theory, estimation and control, distributed algorithms, and reliable programming/simulation. Experience with multi-robot systems, SLAM or traffic data would be relevant, but is not stated as mandatory. (EEE entry requirements, project catalogue)

A concise application portfolio could show a research project with explicit assumptions, algorithm implementation, evaluation and limitations—for example, a small simulation where agents leave and rejoin a communication graph, or a localization/mapping comparison under sparse measurements. Explain what the work demonstrates without implying that the PhD will use the same setup. For a non-UK qualification, prepare transcripts and verify equivalency. Obtain two references in time. Because Home fee status is a hard eligibility condition, resolve it with UCL before submitting if there is any doubt.

Research Direction and Career Outlook

The project’s technical toolkit can translate into robotics and autonomy engineering, localization/SLAM, distributed state estimation and control, autonomous transport, or academic research in control, robotics and machine learning. Skills England’s robotics engineer occupational map describes work spanning specification, design, programming, building and testing robotic systems. A 2025 EngineeringUK update says engineering and technology roles represented 19% of UK jobs. The UK Government’s Digital and Technologies Jobs Plan estimates around 800,000 roles needing to be filled across supported digital and technology sectors between 2025 and 2035. These are broad workforce indicators, not a forecast of robotics vacancies, salary, or this PhD’s graduate destinations; practical systems experience and evidence of implementation remain important. (Robotics engineer occupational map, EngineeringUK workforce update, UK Digital and Technologies Jobs Plan)

Next Actions and Questions to Verify

Before Applying

• Ask UCL Graduate Admissions to assess Home fee status for the 2027/28 start. This project is explicitly Home-only.
• Ask UELA to resolve the £26,022 year-label conflict and confirm the 2027/28 stipend estimate and annual progression.
• Confirm any project-specific degree, English-language or application-material requirements beyond the EEE course page.
• Prepare two referees; aim for receipt by 16 December 2026 and no later than 8 January 2027 at 12:00 GMT.
• Ask Dr Chen for a short discussion of the project scope. A CV and one-page note on research interests and relevant methods could make that conversation concrete.

At Interview

1. What model of network openness is assumed: temporary departure, permanent failure, return with stale data, new agents, or all of these?
2. Which relative measurements and sensors are envisaged, and how will sparse GPS, observability and coordinate-frame ambiguity be handled?
3. What neural implicit representation and distributed learning strategy are being considered? How will an agent’s map be reconciled after it returns?
4. Which traffic datasets, simulation tools, baselines and metrics are planned? Is physical robot or vehicle testing expected?
5. Which theoretical guarantees matter most—convergence, estimation error, robustness to delays, communication cost or some combination?

Before Accepting an Offer

• Get the fee-status decision and Home-rate fee coverage in the formal UCL offer.
• Confirm stipend start-year rate, annual increases, payment duration, RTSG availability and any uncovered project costs.
• Clarify whether the project is full-time only or can accommodate a Home candidate part-time.
• Confirm the four-year registration and thesis-submission arrangements, co-supervisors, equipment/compute access and data agreements.
• Build a 12-month London budget using the actual accommodation choice and expected living costs.

Evidence and Sources

Core Institutional and Programme Sources

• UELA 2027–28 advertised projects catalogue
• UELA studentships: funding, eligibility and process
• UELA application portal
• UCL student fee status
• EEE MPhil/PhD course and entry requirements
• UCL 2027/28 fee schedules
• UCL rankings
• Faculty of Engineering REF 2021
• UCL contact and postal address
• UCL Engineering AI research
• UCL postgraduate living-cost overview
• UCL cost guide

Supervisor and Publication Sources

• Boli Chen’s UCL profile
• Boli Chen publications
• Control and Signal Processing
• Sensors Systems and Circuits
• UCL Robotics Institute
• Fedele, D’Alfonso and Chen, 2025, Automatica
• Zhao et al., 2025, dynamic-graph state estimation
• Sun et al., 2024, reconfigurable-topology path planning
• Chen et al., 2024, online multi-robot coverage planning

Field, Cost and Market Sources

• Decision making in open agent systems
• MACIM: Multi-Agent Collaborative Implicit Mapping
• Di-NeRF: Distributed NeRF for Collaborative Learning
• iS-MAP: Neural Implicit Mapping and Positioning
• Skills England robotics engineer occupational map
• EngineeringUK 2025 workforce update
• UK Digital and Technologies Jobs Plan