Skip to content

Blog

最优控制碎碎念

欢迎来到「最优控制碎碎念」系列!本系列融合施凌老师的 协同感知、Zachary Manchester 的 最优控制、高飞老师的 轨迹规划 以及 Pieter Abbeel 的 深度强化学习 等课程精华,通过严谨的数学推导,构建一个从经典控制到现代强化学习的完整知识体系。我们将从最优感知与控制的基础理论出发,系统探讨模型预测控制、轨迹规划等核心主题,并深入解析深度强化学习在控制领域的前沿进展。

Classic Visual Feature Descriptors

Experiments show that different visual descriptors tend to have similar rates of outliers in feature matching, and the precision differences brought by the application of different feature descriptors in Visual SLAM are negligible.

CoTiMo Planner Report

CoTiMo represents Collision-Free Smooth Path Generation, Time-Optimal Path Parameterization, and Model Predictive Control Planner. It is a comprehensive solution for robotic motion planning.