Mobile 3D Mapping With Tegra K1 Karol Majek karolmajek@gmail.com - - PowerPoint PPT Presentation

▶

Jan 03, 2024 417 likes •636 views

Institute of In f Mathematical Machines Mobile 3D Mapping With Tegra K1 Karol Majek karolmajek@gmail.com Institute of Mathematical Machines www.imm.org.pl Mob obil ile 3D map appin ing Mobile robotic platform Rotating laser

SLIDE 1

Mobile 3D Mapping With Tegra K1

Karol Majek karolmajek@gmail.com Institute of Mathematical Machines www.imm.org.pl

In Institute of f Mathematical Machines

SLIDE 2

Mob

ile 3D map appin ing

Mobile robotic platform
Rotating laser scanner
3D data in stop-scan fashion
Processing using CUDA on robot

SLIDE 3

CUDA in in 3D map apping

Janusz Będkowski – GTC 2012 Now with Jetson TK1

The observations

from robot are in the GPU memory immediately!

More data from the

new sensors

SLIDE 4

CUDA in in 3D map apping

Regular Grid Decomposition
Cube (-1, -1, -1) to (1, 1, 1)
Partitioned into 2n x 2n x 2n

cubes (n=4, 5, …, 9)

Scale the pointcloud

Będkowski Janusz, Andrzej Masłowski, Geert De Cubber. "Real time 3D localization and mapping for USAR robotic application." Industrial Robot: An International Journal 39.5 (2012): 464-474.

SLIDE 5

CUDA in in 3D map apping

CPU CPU OpenMP CUDA Będkowski Janusz, Karol Majek, and Andreas Nüchter. "General purpose computing on graphics processing units for robotic applications." Journal of Software Engineering for Robotics 4.1 (2013): 23-33.

SLIDE 6

3D 3D map apping wit ith Je Jetson TK1

The hardware

Laptop GTX 970m Jetson TK1 CUDA cores 1280 192 Base clock (MHz) 924 + boost 852 Number of multiprocessors 10 1 Architecture (Compute Capability) Maxwell (5.2) Kepler (3.2) Total global memory (MB) 3072 1746 Memory bus 192-bit 64-bit CPU Intel Core i7-4710HQ 2.5GHz 4-Plus-1 quad-core ARM Cortex A15 CPU

SLIDE 7

3D 3D map apping wit ith Je Jetson TK1

13 692 1 10 100 1000 GeForce GTX 970M Jetson TK1 Time in ms

ICP iteration time

Approximately 50 000 points 64 x 64 x 64 grid

SLIDE 8

3D 3D map apping

Data reduction ICP Semantic ICP Classification Normal vector estimation 3D Data

SLIDE 9

3D Data

The 3D datasets: lider.zms.imm.org.pl/

53 scans Odometry and IMU DrRobot Jaguar 4x4 platform Mandala 3D rotating laser unit*

*mandalarobotics.com

SLIDE 10

Data reductio ion

506 064 points 33 530 points 10cm voxel grid

SLIDE 11

Data reductio ion

1,2 1,1 1,1 1,0 0,9 16,6 16,5 15,1 14,5 10,1 0,1 1 10 0,01 0,1 0,5 1 10 Time in seconds Grid size

Data redutcion

GTX 970M Jetson TK1

Input: 2 000 000 points

SLIDE 12

ICP ICP

True NNS Approximate NNS

SLIDE 13

ICP ICP

After 20 ICP iterations

SLIDE 14

Nor

rmal vect

ctor esti timation

Steps:

1. NNS
2. Covariance matrix from NN
3. Principal Component

Analysis (PCA) using Singular Value Decomposition(SVD)

4. Orientation towards the

viewpoint

SLIDE 15

Cla lassific icatio ion

Semantics by point labeling:

Wall (green) – plane
Floor (blue) – plane under robot
Ceiling (black) – plane above robot
Other (red)

SLIDE 16

Cla lassific icatio ion

SLIDE 17

Se Semantic IC ICP

ICP Semantic ICP

SLIDE 18

Con

nclusio

ions

Problems:

TK1 memory is limited to approx 1.7 GB
Low performance compared to modern laptop
Aproximate NNS decreases the final accuracy

Solutions:

Small datasets – data reduction
Use of aproximate NNS instead of true NNS
More iterations of ICP or use of the semantic ICP

SLIDE 19

Ack cknowledgement

This work is done with the support of NCBiR (Polish National Center for Research and Development) project: ”Research of Mobile Spatial Assistance System” Nr: LIDER/036/659/L-4/12/NCBR/2013

SLIDE 20

Than ank you

u for
r you
ur attentio

ion!

Contact information:

e-mail:

karolmajek@gmail.com

LinkedIn:

www.linkedin.com/in/karolmajek

ResearchGate:

www.researchgate.net/profile/Karol_Majek Please complete the Presenter Evaluation sent to you by email or through the GTC Mobile App. Your feedback is important!