Collision Safety

Collision Detection Algorithm

  - Researchers

• • Young-Loul Kim (Ph.D., kimloul@korea.ac.kr), Joon-Hong Kim (M.S., known@korea.ac.kr), Chang-Nho Cho (M.S., sdfzztx2@gmail.com), Sang-Duck Lee (Ph.D., ysdchris@korea.ac.kr)

  - Researcher Period

• • 2008.09. ~

  - Research Background

• • Collision safety must be ensured during human-robot cooperative tasks.

  • To minimize cost, robots must be able to detect collisions without using any additional sensors.

   - Research Objectives

• • Ensure collision safety during human-robot cooperative tasks

  • Detect collisions without using additional sensors/acceleration estimation

  • Quick detection of collisions

    - Research Output

• Collision detection algorithm was developed

• Developed algorithm can detect collisions without need of additional sensors/acceleration estimation

• Developed algorithm can be implemented on any robots with joint torque sensing capability

Collision Reaction Algorithms

    - Researchers

• • Young-Loul Kim (Ph.D., kimloul@korea.ac.kr), Joon-Hong Kim (M.S., known@korea.ac.kr), Chang-Nho Cho (M.S., sdfzztx2@gmail.com), Sang-Duck Lee (Ph.D., ysdchris@korea.ac.kr) 

    - Researcher Period

• • 2008.09. ~

    - Research Backgrounds

• Upon detection of collision, robot must be able to perform appropriate reaction to minimize impact 

• Different reaction strategies for different collision situations

    - Research Objectives

• Develop collision reaction algorithm for various collision situations

    - Research Outputs

• Oscillation mode: 

  • Simulate flexible joints to absorb impact force

  • Easier to resume previous work after reaction

  • For case when user is trapped between robot and wall

• Torque-free motion mode 

  • Simulate robot floating in no gravity, low friction space

  • Robot continues to move away from collision location

  • For case when object continues to move towards robot after collision

• Forced stop mode

  • Move robot at speed proportional to collision force

  • Stop robot immediately after collision

  • For general collision situations

Collision Detection Algorithm to Distinguish between Intended 

Contacts and Unexpected Collisions

    - Researchers

• Young-Loul Kim (Ph.D., kimloul@korea.ac.kr), Joon-Hong Kim (M.S., known@korea.ac.kr), Chang-Nho Cho (M.S., sdfzztx2@gmail.com), Sang-Duck Lee (Ph.D., ysdchris@korea.ac.kr) 

    - Researcher Period

• • 2008.09. ~

    - Research Backgrounds

• During certain tasks, such as teaching-and-playback, intended contact frequency occurs

• Robot should only react to unexpected collisions

    - Research Objectives

• Develop collision detection algorithm that distinguishes intended contacts from unexpected collisions

    - Research Outputs

• Collision detection algorithm to distinguish between intended contacts and unexpected collisions was developed

• Developed algorithm does not require additional sensors/acceleration estimate

• Paper: Joon-Hong Kim, Young-Loul Kim and Jae-Bok Song, Collision Detection Algorithm to Distinguish Between Intended Contacts and Unexpected Collisions, Int. Conf. on Ubiquitous Robots and Ambient Intelligence (URAI 2010), pp. 221-224, 2010.