Robotis Motion Planning

RI16-735
Robotic Motion Planning
NSH 3002, MW10:30-11:50, Fall, 2010

 

 

Main

Syllabus

Lecture

Homework

Student Gallery

Reading Assignment</D\ iv>

 

Personnel:Instructor: Howie Choset, NSH 3211
choset [theat] cs [thedot] cmu [thedot] edu
412-CMU-2495

TA: Glenn Wagner, NSH A403 (temporary)
gswagner [theat] cmu [thedot] edu
412-CMU-

Mailing List:

Secretary: Peggy Martin NSH 3218
pm1e [theat] andrew [thedot] cmu [thedot] edu
412-CMU-7943

Class group email (Email TA to be added/removed)
16735-f10 [theat] googlegroups [thedot] edu

Descriptive Blurb:The robot motion field and its applications have become incredibly broad and theoretically deep at the same time. The goal of the course is to provide an up-to-date foundation in the motion planning field, make the fundamentals of motion planning accessible to the novice and relate low-level implementation to high-level algorithmic concepts. We cover basic path planning algorithms using potential functions, roadmaps and cellular decompositions. We also look at the recent advances in sensor-based implementation and probabalistic techniques, including sample-based roadmaps, rapidly exploring random trees, Kalman filtering, and Bayesian estimation. If time permits, we will study non-linear controls and how it applies to non-holonomic constraints. Click here for more information.

Who should take this class:Advanced undergraduates and graduate students who are new to motion planning.

Text:
H. Choset, K. M. Lynch, S. Hutchinson, G. Kantor, W. Burgard, L. E. Kavraki and S. Thrun,
Principles of Robot Motion: Theory, Algorithms, and Implementations,
MIT Press, Boston, 2005.
Errata
 

Related Robotics Texts:
  • Robot Motion Planning,  Jean-Claude Latombe,  Kluwer Academic Publishers, 1991.
  • Planning Algorithms, Steven M. LaValle.
  • Probabilistic Robotics, S. Thrun, W. Burgard, and D. Fox, MIT press, Cambridge, MA, 2005
  • Where am I?--Sensors and Methods for Mobile Robot Positioning by J. Borenstein , H. R. Everett , and L. Feng
  • Introduction to AI Robotics by Robin Murphy
  • Sensors for Mobile Robots by H. Bart R. Everett
  • Computational Principles of Mobile Robotics by Gregory Dudek and Michael Jenkin
  • Feedback Systems: An Introduction for Scientists and Engineersby Karl J. Astrom and Richard M. Murray
 

Papers:Here is a far-from updated list of papers for your reference

Homework:You are required to create a web page on which you will display your homework assignments. This page should contain a link to each homework's solution. The page for an individual assignment should include a demo of the working program (e.g., gif files, animations), links to source code for your programs (including any necessary makefiles, and a brief explanation of your approach.

Class Project:Propose and implement a robot motion planning project. It can be something related to your research and it must have a motion planning component to it (respect obstacles).   Project proposals will be due at mid-semester (deadlines will be announced soon, and here is a PPT template). As with your homework,  you will create a web page for your project, and it is this web page that will be graded. This page is due 12/12 (two days after the demo) at midnight.

You may find this collection of laser scan data helpful.

Reading Assignment:Based on your interests, we will form groups of one or two to present a paper that go into depth a topic which was covered in the previous week.

Grading
  • 70% Homework
  • 10% Reading presentation
  • 20% Final project

Software:
(No Guarantees)
Please feel free to use software resources that are available in the public domain such as

 

 

### Motion Planning 的配置方法 Motion planning 是机器人学中的一个重要分支,旨在通过一系列的操作生成从初始状态到目标状态的有效路径。以下是对 motion planning 配置的相关说明: #### 1. 自动化规划的基础概念 在自动化规划领域中,已正式研究了如何生成一系列操作以达到目标状态的问题[^1]。这种技术通常用于复杂的环境建模和任务分解。为了实现这一目标,研究人员开发了许多标准化的语言来描述规划问题实例,这些语言可以被现有的规划求解器所理解。 #### 2. 使用 STRIPS 描述问题 STRIPS (Stanford Research Institute Problem Solver) 是一种经典的规划语言,广泛应用于各种场景下的动作序列生成问题。它允许开发者定义一组初始条件、可能的动作以及最终的目标状态。因此,在设置 motion planning 系统时,可以选择 STRIPS 来形式化表达视觉规划问题。 #### 3. 训练与推理效率考量 对于实际应用而言,除了理论框架外还需要考虑计算资源的需求。例如,在某些实验环境中提到的 ACT 模型具有约 80M 参数量,并且能够在单一 GPU 上完成高效训练及快速推理过程[^2]。这意味着如果要部署类似的解决方案,则需确保硬件支持足够的显存容量并评估运行时间是否满足实时性需求。 #### 4. 实际实施步骤概览 尽管不采用具体分步指导方式呈现内容, 还是可以总结几个关键方面帮助构建完整的 motion planner: - **环境建模**: 明确物理空间内的障碍物分布情况以及其他约束因素. - **状态表示**: 设计合理的方式记录当前位置与其他重要变量的状态信息. - **动作集合定义**: 列举所有可行移动选项及其影响效果. - **目标设定**: 清晰指出期望达成的结果形态. ```python def plan_motion(initial_state, goal_state, action_set): """ A simplified function to represent the concept of motion planning Args: initial_state (dict): The starting configuration details. goal_state (dict): Desired end position and orientation specifics. action_set (list): Available actions that can be taken by agent/robot. Returns: list: Sequence of operations leading towards achieving target state. """ current = initial_state.copy() while not is_goal_reached(current, goal_state): best_action = select_best_action(current, action_set) apply_effect(best_action, current) return construct_plan_from_states() ``` 上述伪代码片段展示了基本逻辑流程的一个抽象版本,其中涉及到了判断当前节点是否为目标节点的功能 `is_goal_reached` ,挑选最优下一步行动方案的方法 `select_best_action` 和执行选定行为改变现状的过程 `apply_effect`. ---
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值