ROS 2 Strengths
Massive ecosystem of packages, standardized message passing, simulation integration (Gazebo, Isaac Sim), community support, and industry adoption. ROS 2 is the right choice for teams that need broad sensor integration, existing algorithm implementations, and collaboration with other ROS-based labs.
When Custom Makes Sense
Ultra-low latency requirements (<1ms control loops), minimal dependency footprint (embedded systems), proprietary hardware with non-standard interfaces, and production deployment where ROS overhead is unacceptable. Companies like Boston Dynamics and Figure use custom stacks for their flagship products.
- ROS 2: research labs, multi-sensor integration, rapid prototyping
- Custom: production systems, embedded platforms, latency-critical control
The Hybrid Approach
Many teams use ROS 2 for perception and planning while running a custom real-time control loop for motor commands. This balances ecosystem benefits with control performance. SVRC's OpenArm supports both ROS 2 integration and direct Python/C++ API access for maximum flexibility.