标签：pre roo ken enter lse cloud phi system source
The high level of English is a standard for a top student.
1. Self-trust is the first secret of success.
2. “Movement in a new direction helps you find new cheese.When you move beyond your fear, you feel free.”
3. Understand that life is not a straight line. Life is not a set timeline of milestones. You are allowed to backtrack. You are allowed to figure out what inspires you. You are allowed time, and I think we often forget that.
4. What is it with our need to fast-track relationships? Be alone. Eat alone, take yourself on dates, sleep alone. In the midst of this you will learn about yourself. You will grow, you will figure out what inspires you, you will curate your own dreams, your own beliefs, your own stunning clarity, and when you do meet the person who makes your cells dance, you will be sure of it, because you are sure of yourself.
5. It is common for certain things in life to happen to you. There will be heartbreak, confusion, days where you feel like you aren’t special or purposeful. You cannot let these define you. If you don’t allow yourself to move past what happened, what was felt, you will look at your future with that lens, and nothing will be able to breach that judgment.
6. We get caught up in what our friends are liking, who our significant others are following, and at the end of the day this not only ruins our lives, but it also ruins us.
7. At the end of the day you should be excited to be alive. When you settle for anything less than what you innately desire, you destroy the possibility that lives inside of you. Life and work, and life and love, are not irrespective of each other. They are intrinsically linked. We have to strive to do extraordinary work, we have to strive to find extraordinary love. Only then will we tap into an extraordinarily blissful life.
8. You mean more to me than anything else. ????
9. One day, you‘ll be just a memory for some people. Do your best to be a good one.
10. The longest way must have its close; the gloomiest night will wear on to a morning.
11. Everyone has his own direction to pursue, do not be taken away from the rhythm, to restart a section of the road before you must do a construction of their own hearts, whether the future is better or worse, is just another day. Will there be no end to the end? We are always on the road.
12. Thank you for this gift of a whole new day. I am grateful.
13. Destiny is something we‘ve invented because we can‘t stand the fact that everything that happens is accidental.
14. The more difficult something became ，the more rewarding it was in the end .
15. Stay young, to pack. Whether it is love or youth, in fact, all the way to travel. Do not travel, will be old.
16. Because of the dream, so brave of choice, they only trials and hardships.
17. Life is not a race, but a journey. Race to the end, and travel care about the scenery along the way.
18. Go to different places, see different scenery, know different things, feel different life.
19. If you don‘t go out, you‘ll think this is the world.
20. I have seen many people, I have heard many stories, I have seen the scenery of traveling, and I have learned to grow up.
21. Life is a wonderful journey, on a train that never returns. Go through it with the new and the old. Perhaps this is the fate that a person cannot resist, have you, have me, also have him.
22. A year from now, you are going to wish you started today. So whatever it is you want to do, start doing it. ????
23. I hope everything I‘ve fought for will turn out to be what I want.
Paper
1. Deep learning for object identification in ROS-based mobile robots （https://ieeexplore.ieee.org/document/8394348）
（https://sci-hub.tw/https://ieeexplore.ieee.org/document/8394348）（very useful!）
Abstract:
In this paper, an open source robotic middleware ROS is applied to identify objects with a Raspberry Pi based mobile robot. Faster R-CNN algorithm is considered for enhanced deep learning. The capabilities of robot control and object detection are verified through experimental validations. Based on a low-cost Raspberry Pi control kernel, an easy-to-use interface is developed to integrate front-end and back-end applications. Firstly some ROS packages for the mobile robot are needed to be designed. Then the captured environment information will be provided for objective detection using GPU-accelerated computing. In this paper, Kinect sensor is used to capture images and the deep learning algorithm is implemented based on ROS middleware. All previous mentioned function modules are contained in a cloud service.
Faster R-CNN
Faster R-CNN is efficient for accurate region proposal generation. Faster R-CNN can be easily defined as a system with region proposal network (RPN) and Fast R-CNN. This model has significant improvement compared to R-CNN and Fast R-CNN approaches. The RPN, sharing the convolutional layers with the detection network, leads to real-time performance at inference time and allows an efficient training procedure. This paper compared strength and weakness with three types of R-CNN so in this article implement Faster R-CNN to integration our system and have a proof of concept for our scenario. We survey and compare three papers and select good one to implement and integrate with system. We chose the region-based method Faster R-CNN as a representative architecture among the ones recently proposed in the Deep learning literature for the same task.
The whole scheme of the proposed object detection system with ROS mobile robot.
The scheme diagram of ROS-based image processing.
Implement on Faster R-CNN
In this paper, the object detection system is based on a Faster R-CNN learning network. The Faster R-CNN based object detection is basically composed of the following three essential steps, region proposal, feature extraction, and classification. These three steps will be finally integrated into a deep learning neural network. The Faster R-CNN algorithm can be considered as the combination of region proposal networks and Fast R-CNN system. In this paper, the proposed Faster R-CNN scheme is used for object detection with the integration with a ROS-based mobile robot. Firstly, the Kinect sensor captured images are viewed as the input images filtered in convolutional layers. Then, the region proposal network will provide feature maps. Finally, the classification step predicts object items. In this paper, the Simonyan and Zisserman network model (VGG-16),having 13 shareable convolutional layers, is used as the testbed to fulfill the Faster R-CNN learning algorithm.
Conclusion
In this work, a Raspberry-based mobile robot is constructed for the purse of object recognition. The ROS middleware is applied to develop the functions such as the movement of mobile robot and the image capturing of Kinect sensor. To enhance the recognition results, Faster R-CNN deep learning algorithm is implemented. To reduce processing runtime, Caffe framework is used to build learning network and integrate with GPU acceleration. In addition, the object detection results are involved in a cloud service platform. Experimental results are provided to verify the feasibility of the proposed scheme. Testing results indicate that the desired objects can be identified with high accuracy.
2. A convenient method for tracking color-based object in living video based on ROS and MATLAB/Simulink
Abstract:
A lot of robot platforms are now using the ROS operating system. ROS gave designers a unified and open source platform to design robots. However, the ROS operating system lacks good graphical analysis and operation interfaces. MATLAB has powerful data processing, visual drawing ability and many mature algorithm functions. It is very suitable for algorithm development. At the same time, Simulink has visual interface, which is very convenient for design work. In the process of robot design, it is often encountered to analyze the collected images. Based on ROS and Simulink, this paper introduces a convenient and efficient way to analyze and track moving objects.(the abstract is too simple to understanding the paper)
2018-10-14 星期日
标签：pre roo ken enter lse cloud phi system source
原文地址：https://www.cnblogs.com/sancai16888/p/9784903.html
