Real Time Localization
Real-time localization remains a crucial aspect of the project, significantly contributing to the efficiency and accuracy of detection and tracking in defense applications. This process involves the precise determination of object positions within a given space, in real time, thereby enabling rapid response to dynamic scenarios.
In our context, the real-time localization process has been substantially informed by recent advancements in the field. Specifically, the integration of DeblurSR (Song et al., 2023) and E-CIR (Song et al., 2022) methodologies offer promising avenues for enhancing the localization capabilities.
The DeblurSR method, detailed in section DeblurSR, providnes innovative techniques for super-resolution and deblurring, thereby improving the image clarity required for precise location determination. By applying these advancements, the reliability of our localization process can be significantly augmented.
Similarly, the E-CIR framework, discussed in section E-CIR, introduces novel concepts of information retrieval and object recognition within electronic circuits. These principles can be adapted to improve the computational efficiency of the real-time localization process.
The integration of these methodologies into our real-time localization module (referenced in Milestone 8 in Table 1) demonstrates our commitment to employing state-of-the-art techniques. These inclusions not only reflect the current progress in the project (colored in light green in Table 1) but also outline the direction of our remaining tasks (colored in light blue in Table 1).
| MILESTONE | TASK | START | PLAN END | ACTUAL END |
|---|---|---|---|---|
| Milestone 1 | Project kickoff | 9/30/2020 | 9/30/2020 | 9/30/2020 |
| Milestone 2 | On the Fly sampling | 9/30/2020 | 4/30/2021 | 4/30/2021 |
| Milestone 3 | Camera ISP Code & Paper | 4/30/2021 | 7/30/2021 | 7/30/2021 |
| Milestone 4 | Initial SW Code & demo of DEDRECON | 7/30/2021 | 9/30/2021 | 9/30/2021 |
| Milestone 5 | Signature Detection & Classification | 9/30/2021 | 12/30/2021 | 12/30/2021 |
| Milestone 6 | Integration of IR/EO video streams | 12/30/2021 | 8/30/2022 | 8/30/2022 |
| Milestone 7 | Novel event signature detection | 8/30/2022 | 12/30/2022 | 12/30/2022 |
| Milestone 8 | Real Time localization | 12/30/2022 | 4/30/2023 | 4/30/2023 |
| Milestone 9 | Learning High Fidelity Stable Decision Process Models | 4/30/2023 | 10/30/2023 | |
| Milestone 10 | Coordination with Stakeholder TPOC | 7/30/2023 | 8/30/2023 | |
| Milestone 11 | SW Code for advanced capabilities | 4/30/2023 | 12/30/2023 |
References
-
Song, C., Bajaj, C., & Huang, Q. (2023). DeblurSR: Event-Based Motion Deblurring Under the Spiking Representation. ArXiv Preprint ArXiv:2303.08977.
-
Song, C., Huang, Q., & Bajaj, C. (2022). E-cir: Event-enhanced continuous intensity recovery. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7803–7812.