2026_IR_35_DEI – Multimodal Panoptic Segmentation via Vision-Language Models for 3D Robotic Perception

Italian Ministry of Education, University and Research; Italia; In loco; EUR 30.000 - 38.000; Tempo pieno

Descrizione del lavoro

Dipartimento di Ingegneria dell'Informazione - Università degli studi di Padova invites applications for a research position to advance 3D robotic perception by developing panoptic segmentation algorithms that integrate LiDAR, Radar, and camera data streams. Work is conducted within the EXCEED-AIR seed project on Robust Perception and Situational Awareness.

The ideal candidate will contribute to developing efficient multimodal sensor fusion models, evaluate them on standard datasets and robotic

Mansioni

• Develop efficient Vision-Language Model architectures capable of multimodal sensor fusion under adverse sensing conditions.
• Publish results in prestigious robotics and computer vision journals and/or conferences, and release open-access, FAIR-compliant code and benchmarks.

Descrizione del lavoro

Organisation/Company Dipartimento di Ingegneria dell'Informazione - Università degli studi di Padova Research Field Engineering Researcher Profile Recognised Researcher (R2) Leading Researcher (R4) First Stage Researcher (R1) Established Researcher (R3) Application Deadline 16 Oct 2026 - 13:00 (UTC) Country Italy Type of Contract To be defined Job Status Not Applicable Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

Advancing 3D robotic perception by developing panoptic segmentation algorithms that integrate data streams from LiDAR, Radar, and cameras. The developed algorithms may leverage Vision-Language Models (VLMs) to achieve open-vocabulary segmentation, enabling robots to reliably detect dynamic objects in unstructured environments. Activities will take place within the framework of the EXCEED-AIR collaborative seed project on "Robust Perception and Situational Awareness".

Objectives:

• Develop efficient VLM architectures capable of multimodal sensor fusion under adverse sensing conditions.
• Evaluate models on standard datasets and experimental robotic platforms.
• Publish results in prestigious robotics and computer vision journals and/or conferences, and release open-access, FAIR-compliant code and benchmarks via the EXCEED-AIR Virtual Joint Lab platform.

• OTHER

Eligibility of fellows: country/ies of residence:

• OTHER

Eligibility of fellows: nationality/ies: