Título: Perception and State Estimation for Quadruped Robots in Challenging Scenarios

Autor: João Carlos Virgolino Soares

Seminário da pós-graduação

Horário: às 16h e 0min

Local: on-line

Data: 29/05/2024


Quadruped robots are becoming increasingly significant in both research and industry due to their ability to operate in challenging environments. Compared to other types of robots, they are highly agile and can traverse diverse terrains, overcome obstacles, and carry heavy payloads and tools. Despite these advantages, quadruped robots face several challenges, including real-time perception and robust state estimation. State estimation involves determining the robot's position, orientation, and velocity using onboard sensors such as cameras, laser sensors, and inertial measurement units. With accurate state estimation, the robot can create a representation of its environment, enabling autonomous navigation and obstacle avoidance. These tasks are crucial for the effective field deployment of quadruped robots. However, several factors pose challenges to accurate state estimation, including rough, uneven, and deformable terrain, foot slippage, sensor noise, and moving objects in the scene.

João, a researcher at the Dynamic Legged Systems Lab (DLS) of the Italian Institute of Technology, leads the perception and state estimation team. The lab has over 15 years of experience in quadruped robot research, focusing on the design, locomotion, control, and application of these robots in various fields. This lecture aims to provide a comprehensive understanding of the challenges and advancements in the field of quadruped robotics, with a particular focus on perception and state estimation. We will explore state-of-the-art techniques, discuss ongoing projects, and highlight the contributions of our team.

DLS website:

CV Lattes:

Link do seminário:

Meeting ID: 992 7828 2260
Passcode: 559051