At the time of this review the package was named laser_scan

API review

Proposer: Tully Foote

Present at review:

• List reviewers

Question / concerns / comments

• Location and name suggestions??
  • kwc: naming inconsistency: package is 'laser_scan_utils' but namespace is 'laser_scan'. That implies that the package name should be 'laser_scan'
  • stu: swap 'angle_max' and 'angle_min' in getUnitVectors

  • jml: How does this interact with "TransformListener::transformLaserScanToPointCloud," since this provides similar (but inferior) functionality?

  • vijay: A LaserMedianFilter is defined, but I imagine there could be many more types of laser filters (eg. intensity filter, between-ranges filter)

    • It might be useful to have an abstract LaserFilter type that any laser filter would inherit off of.

• eric: in Median filter, num_ranges and averaging_length are not clearly defined. At least one likely needs re-naming

• eric: in Projector, what is the need for preservative parameter, when the range also defaults to -1?

• brian: Do we intend to keep using NEWMAT data types? Should we instead use ublas or something similar?

Meeting notes

Interface with libtf?

Proposal:

• Move libtf laser projection into here
• Make selection of quality vs. speed selectable
• Pull laser_scan message into this package (Decided against)

Message dependencies:

• Don't want to add deep dependency tree in order to get laser_scan message
• Put laser_scan in robot_msg, its own package, or just depend for messages

Remove preservative

Newmat

• Can keep using
• Don't return Newmat externally

Conclusion

Package status: API partially reviewed. Second review needed.

• Rename to laser_scan

• Move laser_scan into robot_msg

• Move tf laser_scan transform code into laser_scan

• Remove num_ranges, and just allocate when scan received (num_ranges may change)

• Hide getUnitVectors

• Propose and review filter base-class and extensibility

• Review TF integration API