As part of our final year project, we made an autonomous warehouse robot. We wrote a paper about it. This presentation was given as part of paper publishing process in the conference.
You can find presentation here: https://youtu.be/90PcU-07Uko
Low-Cost Autonomous Vehicle for Inventory Movement in Warehouses
1. Low-Cost Autonomous Vehicle for Inventory
Movement in Warehouses
FAISAL ALAM
ASSISTANT PROFESSOR,
COMPUTER
ENGINEERING DEPTT.
ZHCET, AMU
ALIGARH, INDIA
ALAMFAISAL654@GMAIL.COM
ARPIT VARSHNEY
STUDENT, COMPUTER
ENGINEERING
ZHCET, AMU
ALIGARH, INDIA
VARSHNEYARPIT158@GMAIL.COM
KHAN SAAD BIN HASAN
STUDENT, COMPUTER ENGINEERING
ZHCET, AMU
ALIGARH, INDIA
KHANSAADBINHASAN@GMAIL.COM
2. Project Objectives
• Automation can help Warehouses become more efficient,
productive, robust, and cost-effective.
• A variety of Autonomous Robots are used in large
warehouses which can be expensive.
• We intend to build cost-effective Autonomous Robots for
small and medium scale Warehouses with reasonable
accuracy.
3. Other approaches
• Amazon[1] robots are monitored by a central administrator.
The robots navigate by using QR Codes, which are pasted on
the ground.
• Diop et al,[2] Attached two cameras on the robot. The authors
developed the algorithm to process the images from cameras
to achieve robot position and orientation.
• Deepu et al,[3] Demonstrates the approach of Robot
Navigation using Laser Light and the single-camera attached
to the robot.
6. Our approach
• The Source and Destination are provided by the user in the
environment visible to camera.
• A path is made from source to destination using the A-
star Algorithm.
• A colored marker is placed on the robot which is used to
estimate the robot’s location and the heading.
7. Our approach
• A line is drawn from the yellow to
the purple marker which gives the
Heading.
• The decision to move the next step
is taken using the heading of the
robot, the robot’s current location,
and the next coordinate in the path
to follow.
8. Evaluation Parameters
• Cost and Accuracy: There is a trade-off between the accuracy of
the robot and the cost required to build it.
• Scalability: Using an overhead camera for tracking, robots can
easily be scaled up, to track and guide multiple robots .
• Backup system: Redundancy should be available that could be
used in case of any failure with the primary system.
• Path Following and Reaching Destination: To track the
deviation while the robot is traversing the estimated path and
check whether the robot can reach the destination or not.
9. Evaluation Methodology
• We Evaluate the results on the success rate of the robot in
reaching the destination and the deviation in the path to reach
the destination.
• We have considered that if the distance of the robot and
destination is less than 20cm or 100px, then the robot is on the
destination point.
• We draw the path that the robot should move on and the path
that the robot moved on. We compare these paths to measure
how close the robot was to the destination.
10. Evaluation Results
• The computed path is shown in orange color and the actual
path traversed by the robot is shown using the blue color.
11. Evaluation Results
• Out of 10 times, The robot was able to reach the destination 6
times. Hence, the accuracy of our model should be close to 60%.
• The robot is built with cheap hardware which is easily available
making our robot very affordable.
• The robot also uses an overhead camera which facilitates greater
scalability.
12. Conclusion and Further work
• We have built a cheap and fairly accurate robot. It is versatile
and can work in varied environments. Given the backup
system and an overhead camera, the robot is fairly secure and
scalable.
• The scope of the work can be enhanced in the future by
employing novel ways of increasing accuracy with little
changes in cost.
13. References
[1] D'Andrea, Raffaello. "Guest editorial: A revolution in the warehouse: A retrospective on kiva
systems and the grand challenges ahead." IEEE Transactions on Automation Science and Engineering
9.4 (2012): 638-639.
[2] Diop, M., Ong, L. Y., Lim, T. S., & Lim, C. H. (2016, April). A computer vision-aided motion-
sensing algorithm for mobile robot's indoor navigation. In 2016 IEEE 14th International Workshop
on Advanced Motion Control (AMC) (pp. 400-405). IEEE.
[3] Deepu, R., B. Honnaraju, and S. Murali. "Path generation for robot navigation using a single
camera." Procedia Computer Science 46 (2015): 1425-1432.