Trajectory interpolation: prediction of missing points in movement data using long short-term memory

Zijian Wan¹, Somayeh Dodge¹

¹Department of Geography, University of California Santa Barbara, USA

Abstract

Movement tracking devices usually sample the location of entities at mostly regular intervals as sequences of timestamped locations, named trajectories. However, many factors (e.g., battery outage, signal loss) may lead to the interruption of tracking data recording, which results in missing data in trajectory datasets. These missing data, termed gaps in this study, need to be dealt with before further analysis. This study proposes a new trajectory interpolation model that leverages a generative adversarial network (GAN) architecture to predict missing trajectory points. In this model, two modules, namely a generator and a discriminator, work against each other during the training process. Long short-term memory (LSTM) layers are used in both the generator and the discriminator of the constructed GAN to process the trajectory data. In the generator, an encoder-decoder structure is leveraged, in which the encoder reads the two observed trajectory segments surrounding a gap, and then the decoder interpolates what is missing in the middle according to the information provided by the encoder. During the training process, the generator aims at generating interpolation results that are close enough to the ground truth, which might fool the discriminator, while the discriminator aims at not getting fooled by the generator. Following the architecture of InfoGAN, we add a latent code in addition to the noise to not only avoid GAN mode collapsing but it also helps to deal with multi-modality in trajectory interpolation. In this paper, we describe the model and assess the proposed model using a real-world GPS trajectory dataset.

Keywords

Trajectory interpolation; movement modeling; generative adversarial network (GAN); long short-term memory (LSTM); GPS trajectories

Session info

John Odland Award (SAM student paper competition)

The Spatial Analysis and Modeling (SAM) Specialty Group of the American Association of Geographers (AAG) is sponsoring the John Odland Award (SAM student paper competition) at the 2023 AAG Meeting in Denver, Colorado. The prizes will total a minimum of $1,000 and up to $1,500.

The competition is open to active undergraduate and graduate students who have previously not won the award. Papers may be of a theoretical or applied nature. They will be judged on the following criteria:

Potential contribution to the use of mathematical models, statistical techniques, and other technological and computational approaches for analyzing spatial phenomena in any subfield of geography; Appropriate and sound use of methodology; Originality; Organization and written composition of the paper; Quality of oral presentation.

Students wishing to enter the competition should submit the title and 1000-1500 word extended abstract (using the required format [download here]) of their paper to SAM Board Member Taylor Oshan (toshan@umd.edu) after registering to attend the 2023 AAG Annual meeting. Students should also include the PIN from their registration so that SAM can coordinate with the AAG to place competition papers into special sessions. The deadline to enter the competition is November 7, 2022. The committee will review the abstracts and let each applicant know if s/he is selected for participating in the paper competition session by December 2, 2022. All invited students entering the competition must submit their completed papers in .doc or .pdf format to the same e-mail address (toshan@umd.edu) by January 20, 2023. Due to the hybrid format of the 2023 Annual Meeting, and to ensure fairness and inclusion, students entering the competition must also submit a video recording of their presentation by February 3, 2023. Recordings may not be longer than 15 minutes. Late submissions of papers and recordings will not be accepted. Competitors must participate in the conference sessions, whether virtually or in-person, and failure to do so will result in disqualification. To encourage in-person participation, SAM has student travel awards available and there will be a special cash prize for the best in-person presentation that is considered and awarded parallel to the main competition.

The paper must be based on research primarily conducted while the student was at an accredited university. Coauthored papers are accepted as long as the student is the primary author of the manuscript. Each entrant must submit a statement with their completed paper from a university faculty member, preferably their undergraduate or graduate advisor, certifying the role of the student as the lead contributor in completing the paper by January 20, 2023. If the committee does not receive a letter from a faculty supervisor, the candidate is automatically disqualified from the competition. The same paper cannot be submitted for multiple competitions and may not be submitted for multiple years.

The title page of the submitted paper should include the name, current affiliation, mailing address, and e-mail address of the entrant and coauthors if any. The following page should include only the title of the paper and an abstract. No identifying information should appear anywhere else than on the title page of the paper. Papers should be no longer than 35 double-spaced pages, including tables, figures and references.

A panel of judges will review the papers before the AAG meeting. The judges will also evaluate recordings of student presentations. The main competition (1st place, 2nd place, etc.) will NOT be based on the evaluation of the live presentations, though participation as a presenter (Either virtually or in-person) during live sessions on the day of the competition is required. All those that register, attend, and present in-person at the conference will automatically be considered for a special additional cash prize for the best in-person presentation that will be awarded independently of the main competition. The winners will be announced at the SAM specialty group business meeting and are invited to the AAG Awards Luncheon. All decisions, including the possibility of not awarding a prize, are final.