| June 2024: Experiments with Qualisys Motion Capture System | |
|---|---|
|
Participants: Nicolas Chleq, Pierre Joyet, Louis Verduci, Quentin Louvel, Jon Aztiria-Oiartzabal, Pardeep Kumar, Mohamed-Malek Aifa, Andres Gomez-Hernandez, Clara Thomas Context and problems: The goal was to test the efficiency of various control and perception algorithms on a drone in an immersive indoor environment. The main challenge was to create accurate models and simulations of drone dynamics and behaviors to enhance navigation, control, and obstacle avoidance. Qualisys provides advanced motion capture systems using high-resolution cameras and reflective markers to accurately record movements for drone applications. Their software reconstructs these movements in 3D, crucial for precise navigation, obstacle avoidance, and aerial maneuvers. C-Motion develops software for biomechanical modeling that utilizes Qualisys data to analyze movement dynamics, aiding in optimizing drone flight paths, enhancing navigation algorithms, and improving overall drone performance |
|
| March 2024: Detection and Tracking of Moving Objects with an Event Based Camera – Thomas Campagnolo | |
|
Participants: Thomas Campagnolo, Carmine Tommaso Recchiuto, Philippe Martinet and Ezio Malis Context and problems: In the context of Autonomous Vehicles perception, the University of Genoa and the ACENTAURI team from Inria collaborated to develop a system capable of detecting and tracking only the moving objects around the vehicle’s surroundings, using an event-based camera. In consequence, both co-supervise a master thesis centered on: Detection and Tracking of Moving Objects with an Event-based Camera on an Autonomous Car. The main challenge in perception is achieving high precision and accuracy to ensure the system’s affidability and integrity. Contributions: To address the perception problems using an event-based camera, this method uses a global minimization to perform camera motion compensation, essential for stabilizing the scene distorted by the sensor’s ego-motion. The implemented detection phase involves clustering the moving objects present in the event frames using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm, with hyperparameters determined dynamically based on the distance to the k-th nearest neighbor (KNN) graph and the percentile of these distances, influencing the cluster size and identification of cluster points, to group spatially and temporally related events that probably belong to the same moving object. The Simple Online and Real-time Tracking (SORT) algorithm is used for the tracking stage, detailing the prediction model of the moving objects, identity association, and state update processes. |
|
| October 2023: Towards autonomous robot navigation in human populated environment using an Universal SFM and parametrized MPC – Enrico Fiasché | |
|
Participants: Enrico Fiasché, Philippe Martinet and Ezio Malis Context and problems: A train station simulation where a robot assists a blind person and guides him to the correct platform. In the simulation, the robot must avoid collisions with obstacles and other pedestrians while completing its task. Contributions: Autonomous mobile robot navigation in a human populated and encumbered environment is recognized as a hard problem to be solved in real-time. Most of the time, robots face the so-called ‘Freezing Robot Problem’, that occurs when the robot stops because no feasible and safe motion can be found. In order to provide to the robot the capability of proactive navigation, in this work we generalize the classical Social Force Model into a Universal Social Force Model (USFM) that attributes to any object surrounding the robot (humans, robots, obstacles) a social behavior. Nonlinear Model Predictive Control (MPC) can be used to solve the autonomous navigation problem since it can take into account all the possible constraints coming from the interaction model between the robot and the different surrounding objects. However, to be effective, MPC requires a sufficiently large prediction horizon, which generally implies a high computational cost. In order to considerably reduce the computational cost, we propose a new control parametrisation based on Thin Plate Spline Radial Basis Functions that allow us to have a large prediction horizon with fewer parameters. The global control framework is validated in simulation with virtual pedestrians, and in real world environments. |
|
|
|
| September 2023: Ma thèse au Cerema – Diego Navarro | |
|
Participants: Diego Navarro, Ezio Malis, Rahael Antoine, Philippe Martinet, Cyrille Fauchard Context and problems: In the context of the ROAD-AI challenge, the ENDSUM team from Cerema and the ACENTAURI team from Inria are collaborating to develop a system capable of conducting radar inspections in close proximity to the surface of structures. In consequence, both teams co-supervise a doctoral thesis centered in: Precise localization and control of a collaborative multi-robot system for infrastructure inspection. The main challenge in localization is achieving high precision to ensure the system’s integrity. Additionally, there are issues related to potential loss of GPS signal and a lack of meaningful information during close-range flights. Contributions: To address the localization problems, the system is expected to incorporate heterogeneous sensor fusion, combined with a V-SLAM (Visual Simultaneous Localization and Mapping) framework. This parallel approach will create a dense representation of the environment for navigation tasks. Moreover, the control of the agent will utilize model predictive control and state observers to compensate for aerodynamic perturbations specific to the user case. |
|
| December 2019: Mapping and Autonomous navigation close to Sophia-Arena | |
|
Participants: J. Thomas, P. Martinet, S. Dominguez Quijada (LS2N) |
|
|
|
| January 2019: Mapping and Localization in Sophia-Antipolis | |
|
Participants: J. Thomas, P. Martinet, S. Dominguez Quijada (LS2N) |
|
|
|
| January 2019: Autonomous navigation at INRIA Sophia-Antipolis | |
|
Participants: J. Thomas, P. Martinet, S. Dominguez Quijada (LS2N) |
|
|
|
| 2019: Car Platooning at Ecole Centrale of Nantes | |
|
Participants: A. Khalifa (LS2N), O. Kermorgant (LS2N), S. Dominguez Quijada (LS2N), P. Martinet (Inria) S. Dominguez Quijada, B. Khomutenko, G. Garcia, P. Martinet, A. Khalifa, O. Kermorgant, S. Dominguez Quijada, P. Martinet, “ Vehicles Platooning in Urban Environment: Consensus-based Longitudinal Control with Limited Communications Capabilities”, ICARCV’18, pp. 809-814, Singapore, Singapore, November 18-21th, 2018 |
|
| 2019: Autonomous Parking at ECN site | |
|
Participants: D. Perez Morales (LS2N), O. Kermorgant (LS2N),S. Dominguez Quijada (LS2N) P. Martinet (Inria) Context and problems: In the framework of VALET ANR porject, Autonomous parking problem is adressed. Different problems have been adressed: relative localization regarding an empty slot, perfoming maneuvers in reduced and dynamic environment. Collaboration between CHORALE and LS2N/ARMEN. D. Perez Morales, O. Kermorgant, S. Dominguez Quijada, P. Martinet, “ Autonomous Perpendicular And Parallel Parking Using Multi-Sensor Based Control”, 9th Workshop on Planning, Perception and Navigation for Intelligent Vehicles, IROS17-PPNIV’17, pp. 38-44, Vancouver, Canada, September 24th, 2017 |
|
