代做UFME7R-15-M Robot learning and teleoperation 2024/25代做Statistics统计

- 首页 >> Java编程

Assessment Brief

Submission details

Module title:                             Robot learning and teleoperation

Module code:                           UFME7R-15-M

Assessment title:                     Report of robot learning and teleoperation

Assessment type:                    Written report

Assessment weighting:          50% of total module mark

Size or length of assessment: maximum 1500 words

New Assessment Content Limit Policy for the University (sharepoint.com)

Module learning outcomes assessed by this task:

1. Apply robot learning algorithms to teach robots new skills and behaviours, enabling them to adapt and generalize from training data.

2. Design and implement teleoperation systems for safe and efficient human-robot interaction, with considerations of control architectures, communication protocols, and user interfaces.

Submission and feedback dates

Submission deadline:             Before 14:00 on 1st  May 2025

This assessment is eligible for 48 hour late submission window

Submission format:

Submit your MLX file and a PDF render of it. This should be done with outputs, so you must Run the whole file successfully to display your results and graphs

Marks and Feedback due on: 30th  May

N.B. all times are 24-hour clock, current local time (at time of submission) in the UK

Marks and Feedback will be provided via:

Blackboard

Completing your assessment

What am I required to do on this assessment?

The assessment task entails producing a technical report that details the integration of the acquired robot learning methodology and teleoperation. The report must be submitted prior to the deadline, following the instruction provided by the module leader. Please complete the report under the guidance with the following tasks and steps:

Task   1:   Utilize    Dynamic   Movement   Primitives   (DMPs)   discussed   in   the    Robot   Skill Generalisation, Gaussian  Mixture  Models (GMM) in  Humane Skill  Encoding, and Gaussian Mixture  Regression  (GMR)  in  Robot  Skill  Regression,  to   devise  an  enhanced  Dynamic Movement Primitives for single-arm robot manipulation. During the development of the new algorithm, you can refer to the paper “DMP and GMR based Teaching by Demonstration for a KUKA LBR Robot” introduced in the reading week lecture. Write an introduction about the calculation  steps  based  on  your  understanding  of  this  paper  into  your  own  report.  The relevant code is provided on the Blackboard. (500 words, 10 Marks)

Task 2: Implement your improved DMP method using the following steps (50 marks)

1) Employ DMPs to compute the values of nonlinear term (the forcing term) of the inputs. This step can be realized by using the code from original Dynamic Movement Primitives. (10 Marks)

2) Utilize Gaussian Mixture Models (GMM) and Gaussian Mixture Regression (GMR) instead of conventional nonlinear terms in DMP to carry out a nonlinear regression for the calculation results in step 1. (15 Marks)

3) Once the new trajectory has been learned and generalized, save the data in a .mat format to the workspace. (5 Marks)

4) During teleoperation simulation, load the saved data as inputs for the teleoperation system to facilitate further simulation. Please complete simulations and make a screenshot of the simulation results with the scopes to illustrate the simulation results. (15 Marks)

Task  3:   Evaluate  the   proposed   enhanced   Dynamic   Movement   Primitives   method   by comparing  the  differences  between  the  original  Dynamic  Movement  Primitives  and  the enhanced method in Cartesian space coordinates. The successful results should demonstrate comparative fitting errors. Complete the comparison and analysis for the results and write them in the report. (500 words, 15 Marks)

Task 4: Save the outputs of robot learning and load them into the teleoperation model in Simulink as inputs on the leader controller side to guide the movement of the follower. By designing PD or other controllers on both the leader and follower, the expected outcome is for the follower's desired motions to mirror those on the leader side. Consequently, save and plot the positions on both the leader and follower sides, and compare the disparities between them. The comparison and the control diagram built in Simulink should also be included in the report. (500 words, 15 Marks)

Complete the aforementioned tasks and document them in your report. In the report, you should describe the complete process adopted in each task and make a comparison of the improved method against the general DMPs. You should also illustrate how the improved method can deal with the cooperative robot manipulation. The introduction of DMPs, the methodology, analysis and conclusions must be clearly and logically presented in the report.

(10 Marks)

Where should I start?

The  coursework  will  commence  with  a  reading  lecture  introducing  the  paper  'Dynamic Movement  Primitives  for  Cooperative  Manipulation  and  Synchronized  Motions,'  wherein participants will learn how to apply the acquired method in actual research. Subsequently, please consult the codes provided by the module leader, located in the Blackboard folder, to gain  a  comprehensive  understanding.  Following  this,  proceed  to  enhance  the  algorithm introduced in the assessment section to meet the aforementioned tasks. Robot learning is employed to generate robot movement, and these motions are subsequently used as inputs for teleoperation. Please refer to the detailed tasks and complete this report containing the essential components presented in the Criteria.

What do I need to do to pass?

Pass(2:2) >50%: Load the data to realize motion planning using general dynamic movement primitives without improvement. Use the obtained data as inputs for teleoperation and implement PD control to achieve motion synchronization between the leader and followers.

Pass(2:1) >60%: Load the data to achieve motion planning employing improved dynamic movement primitives as per the provided reference. Utilize the acquired data for teleoperation inputs and employ PD control to ensure motion synchronization among the leader and followers. The report contains essential sections and provides sufficient evidence.

Distinct >70%: Load the data and employ improved dynamic movement primitives based on the provided reference for motion planning. Utilize basic control methods except for the PD control to achieve motion synchronization among the leader and followers during teleoperation. Conduct a detailed analysis of robot execution. The report's essential sections are clearly presented.

How do I achieve high marks in this assessment?

Please refer to the above criteria to achieve high marks.

How does the learning and teaching relate to the assessment?

All the lectures and tutorials covered all the sections relating to this coursework e.g., Linear regression,  Human  skill  encoding,  Robot  skill  regression  and  Skill  generalisation  with Dynamics to teach robots new skills and behaviours. The generalized trajectories are utilized for  robot  teleoperation.  Hence,  students  are   required  to   integrate  robot  learning  and teleoperation to fulfil the assessment.

What additional resources may help me complete this assessment?

The uploaded code and slides provide in the Blackboard Folder.

What do I do if I am concerned about completing this assessment?

It is recommended that you review all of the relevant materials on Blackboard. You can also speak to your module leader for advice and guidance.

UWE Bristol offer a range of Assessment Support Options that you can explore through this link, and both Academic Support and Wellbeing Support are available.

For further information, please see the Student study essentials.

How do I avoid an Assessment Offence on this module?

Use the support above if you feel unable to submit your own work for this module.

1. Complete the assessment following the guidance of this module

2. Writing in the correct form. following the guidance of this template

3. Avoid copying code and context from the internet or using AI Tools.

4. Ensure the complement of all the task work and present the improvement to the original algorithm

5. Complete the report.



站长地图