Salem Robotics (YC S26) launches software for industrial inspection robots

The founders said they came to Salem through robotics research at UT Austin and a combined 15 years working in nuclear, including about 10 years developing and deploying autonomous robots at Los Alamos National Laboratory. Over the last five years, they kept running into the same gap: robot hardware had become very capable, but making a robot carry out a complete industrial procedure still required a surprising amount of robotics work and manual intervention.

The part that interested them most is manipulation. A nuclear contamination survey, for example, requires taking a 'smear,' wiping a defined area of a surface to check for removable radioactive contamination. In an oil, gas, or chemical facility, an LDAR (leak detection and repair) inspection can require moving a detector around a particular valve, flange, or connection. Other inspections require positioning an instrument at a precise location and orientation relative to a pipe or piece of equipment.

These tasks are easy to describe in words but difficult to make a robot do reliably, the founders explained. A probe might need to remain normal to a surface throughout a path, stay within a narrow offset from a pipe, or trace a region while maintaining a particular end-effector orientation. The planner has to find a feasible motion while respecting the task geometry, manipulator kinematics, joint limits, collisions, and the environment.

Salem Robotics works down to joint-level control for those interactions. One problem they said they spent a lot of time on is generating constrained manipulation plans quickly enough that they can be based on the geometry the robot actually observes, instead of requiring someone to carefully author a trajectory for every individual surface, valve, or flange.

The physical world makes this difficult, they noted. A few centimeters of error may not matter when navigating a hallway, but it matters if a sensor is supposed to remain normal to a curved surface. Executing a trajectory successfully does not necessarily mean the inspection worked; the detector could be misaligned, contact could be wrong, the geometry could differ from the model, or the measurement itself could be invalid. The company focuses on closing the loop around the inspection result, not just whether the arm reached the commanded pose.

Their approach combines AI and classical robotics. Much of the field is moving toward end-to-end learned systems, particularly around humanoids, but working in safety-critical environments has made the founders appreciate the relevance of classical approaches when explicit constraints, predictable behavior, and theoretical guarantees are needed. They use AI for semantic understanding and flexibility, such as interpreting less structured information or understanding what in an unfamiliar scene is relevant to a procedure. Once the system knows the physical interaction, they prefer explicit geometry, planning, optimization, and control.

The company is hardware agnostic. They do not expect one robot to be the best platform forever, and facilities already own different hardware. They describe an inspection in terms of what needs to happen and map that onto the capabilities of the right robot for the job. This is also why they believe every useful robot application need not require building a new robot; a domain-specific application layer can sit on top of existing platforms from companies like Boston Dynamics.

The founders said they were surprised by how manual many inspection workflows remain. In sophisticated nuclear and industrial sites, people still physically walk survey routes, take measurements one at a time, visually inspect equipment, record results manually, and sometimes make judgments based on things like how a component sounds. The company is starting with radiological inspection in nuclear because that is the industry they know well.